Rainstorm early warning method and system based on wireless network, terminal and storage medium

Abstract

The invention relates to the technical field of meteorological disaster monitoring and early warning, in particular to a rainstorm early warning method and system based on a wireless network, a terminal and a storage medium, and the method comprises the steps: collecting multi-source data through distributed sensor nodes; preprocessing the multi-source data to obtain real-time data; constructing a spatial-temporal feature extraction model; generating a dynamic threshold function, performing weighted fusion on the dynamic threshold function and a preset static threshold, and outputting a graded early warning threshold; judging whether the current accumulated water depth acceleration exceeds a gradient critical value of a graded early warning threshold value or not; and if so, outputting a rainstorm early warning level. The method has the advantages that the problem that a static threshold mechanism cannot dynamically adapt to rainstorm disaster risks in a complex environment is solved, and the accuracy and practicability of the rainstorm early warning system are improved.

Description

Technical Field

[0001] The present application relates to the technical field of meteorological disaster monitoring and early warning, and in particular to a wireless network-based rainstorm early warning method, system, terminal and storage medium. Background Art

[0002] A wireless network-based rainstorm warning method primarily uses wireless sensor networks, the Internet of Things, and communication technologies to collect key parameters such as rainfall and water depth in real time. Combined with intelligent analysis models, it enables real-time warning and emergency response for rainstorm disasters. This method, through multi-dimensional data fusion and dynamic threshold judgment, addresses the challenges of traditional warning systems, such as their reliance on manual experience and delayed response. This provides a scientific basis for urban waterlogging prevention, traffic scheduling, and evacuation.

[0003] In related technologies, at the data collection layer, sensors are deployed in areas such as cellars, streetlights, and rivers, and the status of manhole covers is detected by using cellar pressure sensors, and rainfall monitoring modules are integrated into streetlights to monitor environmental parameters in real time. At the data transmission layer, a three-layer tree topology is constructed through the ZigBee network, and multi-node networking and data transmission are achieved by using wireless mesh networks to achieve centerless self-organizing networks. Alarms are triggered based on static thresholds, such as activating sound and light alarms or pushing warning information through APP when the water level exceeds the preset value. Regarding the related technologies mentioned above, the static threshold mechanism cannot dynamically adapt to the risk of rainstorm disasters in complex environments. In sudden extreme rainfall or low-lying areas with complex terrain, fixed thresholds are prone to false alarms (such as short-term heavy rainfall that does not reach the threshold but has caused waterlogging) or missed alarms (such as long-term cumulative rainfall exceeding the threshold but the system does not respond in time). In addition, traditional methods lack real-time fusion analysis of multi-source data (such as weather forecasts, drainage network topology, and historical waterlogging records), making it difficult to achieve dynamic assessment and graded warning of disaster risks, which restricts the accuracy and practicality of rainstorm warning systems. Summary of the Invention

[0004] In order to improve the problem that the static threshold mechanism cannot dynamically adapt to the risk of rainstorm disasters in complex environments and improve the accuracy and practicality of the rainstorm warning system, the present application provides a rainstorm warning method, system, terminal and storage medium based on a wireless network.

[0005] In a first aspect, the present application provides a rainstorm warning method based on a wireless network, which adopts the following technical solutions: A rainstorm early warning method based on a wireless network, comprising: Collect multi-source data through distributed sensor nodes, including environmental data, drainage network flow data, and weather radar data; Preprocess multi-source data to obtain real-time data; Correlate and match real-time data with historical databases to build a spatiotemporal feature extraction model; Generate a dynamic threshold function based on the spatiotemporal feature extraction model, and perform weighted fusion of the dynamic threshold function with the preset static threshold to output a graded warning threshold; During real-time monitoring, determine whether the current water depth growth rate exceeds the gradient critical value of the graded warning threshold; If it exceeds, the heavy rain warning level will be output based on whether the current water depth growth rate exceeds the gradient critical value of the graded warning threshold.

[0006] By employing this technical solution, a dynamic threshold function is generated using a spatiotemporal feature extraction model. This function's parameters are dynamically adjusted by combining historical databases with real-time data. Dynamic and static thresholds are weighted and fused to produce graded warning thresholds, avoiding the limitations of a single fixed threshold. When short bursts of heavy rainfall fail to reach the fixed threshold but already cause urban flooding, the dynamic threshold can trigger timely warnings by analyzing the increasing rate of accumulated water depth and environmental characteristics in real time, reducing false alarm rates. In scenarios where long-term cumulative rainfall exceeds the threshold but the system fails to respond, the dynamic threshold model leverages historical waterlogging records and current trend forecasts to proactively identify risks and reduce missed alarms. By leveraging a dynamic threshold mechanism, multi-source data fusion, real-time monitoring and rapid response, intelligent model optimization, and tiered warning linkage, the rainstorm warning system is enhanced in terms of accuracy, real-time performance, and practicality. Compared to traditional methods, this significantly reduces false alarm and missed alarm rates, significantly speeding up emergency response times several times and providing a scientific basis for urban waterlogging prevention, traffic scheduling, and personnel evacuation.

[0007] Furthermore, the deployment method of the distributed sensor nodes includes: Obtain key deployment points and deployment density based on the geographical location and topographic features of the monitoring area; Determine sensor categories based on key deployment points; After deploying sensors in the monitoring area according to key deployment points, deployment density, and sensor types, calculate the actual data fluctuation value of the returned data based on the return data of the corresponding sensors deployed in the monitoring area; Determine whether the actual data fluctuation value is greater than the preset fluctuation threshold; If yes, correct the key deployment points.

[0008] By employing this technical solution, during the deployment of distributed sensor nodes, the actual data fluctuation of the returned data is calculated to determine the appropriateness of the current deployment points. If the fluctuation value exceeds a preset threshold, key deployment points are dynamically adjusted to avoid monitoring blind spots or redundancies caused by improper initial deployment. Sensor density and types are dynamically adjusted based on the geographical location of the monitoring area (such as urban flood-prone areas and landslide-prone areas in mountainous areas) and topographic features (such as low-lying areas and river bends). Dynamic adjustment of deployment points effectively improves monitoring coverage, ensuring comprehensive coverage of high-risk areas. Fluctuation value analysis eliminates redundant nodes, optimizes sensor deployment density, and reduces hardware costs and energy consumption. This dynamic adjustment mechanism enables the system to adapt to extreme weather (such as heavy rain and flooding) or environmental changes (such as river diversion and urban construction), extending the effectiveness of the deployment plan.

[0009] Furthermore, the method further comprises: Obtain sensor data and historical climate information corresponding to the monitoring area in the historical database; Establish a correlation function between sensor data and historical climate information; Obtain corresponding weather warning information; Anchor trusted sensors based on weather warning information and correlation functions; A sampling frequency update instruction is sent to the trusted sensor, where the sampling frequency update instruction is used to instruct the trusted sensor to increase the sampling frequency.

[0010] By employing this technical solution, we establish a correlation function by analyzing sensor data from historical databases, such as rainfall, temperature, humidity, and wind speed, with historical climate information. This function quantifies the reliability of data from different sensors under specific climate conditions. By filtering out anomalous data from untrusted sensors, we can reduce the impact of these data on the early warning model and improve the accuracy of graded early warnings. During peak rainfall periods (e.g., rainfall >20mm in 10 minutes), the sampling frequency of trusted sensors automatically increases to 20 times per minute to ensure that critical data is not lost. During periods of stable rainfall, the sampling frequency gradually returns to normal, conserving network resources.

[0011] Furthermore, the method further comprises: Obtain historical location information and current location information of the target sensor; When the displacement between the historical position information and the current position information is greater than a preset displacement threshold, defining sensors near the target sensor as neighboring sensors; Obtaining a target data value of a target sensor and adjacent data values ​​of adjacent sensors; Determine whether the target data value is valid based on the adjacent data values; If yes, output the target data; If not, the target data is eliminated.

[0012] By employing this technical solution, the target sensor's historical location information is compared with its current location information, the displacement is calculated, and compared with a preset threshold, dynamically identifying whether the sensor has experienced significant displacement, such as being swept away by floods. When the displacement exceeds the threshold, sensors near the target sensor are defined as neighboring sensors, and the data from these neighboring sensors is used to cross-validate the validity of the target sensor's data. Sensor data that has been swept away or significantly displaced is promptly identified and eliminated, reducing the proportion of invalid data and avoiding false positives or omissions caused by erroneous data. Real-time monitoring of changes in the sensor's physical location ensures the system can quickly respond to equipment anomalies such as flood damage and vandalism, shortening the time it takes to detect equipment failures. Through the data verification mechanism of neighboring sensors, invalid data is eliminated, improving the completeness of valid data in the target area.

[0013] Furthermore, after outputting the target data, the method further includes: Update the spatiotemporal feature extraction model based on the target data and neighboring data values; Generate a dynamic threshold update function based on the updated spatiotemporal feature extraction model; The output value of the dynamic threshold update function is weightedly fused with the graded warning threshold to output the updated graded warning threshold.

[0014] By adopting the above technical solution, once the target sensor data is determined to be valid, it is input into the spatiotemporal feature extraction model along with the data values ​​from neighboring sensors, and the model parameters are updated to ensure that the model can quickly adapt to new data while retaining the characteristic information of historical data. The valid data from the target sensor is fully incorporated into model training, improving data utilization and reducing data waste. By dynamically updating the spatiotemporal feature extraction model, the model's prediction error rate is reduced, especially in extreme weather scenarios such as peak rainstorms. The model's feature capture capability is significantly enhanced, and the residual value of the target sensor data is fully utilized during the short validity period before it is eliminated, avoiding potential information loss due to data elimination.

[0015] Furthermore, the method for switching the information transmission path includes: When the rainstorm warning level is greater than or equal to the preset warning level, the communication protocol switching instruction is triggered to control the transmission link of multi-source data to switch to the high-priority communication protocol; After the transmission link switches to a high-priority communication protocol, the quality of multi-source data return is tested; When the backhaul quality falls below a preset quality threshold, a relay node is deployed to compensate for the transmission link; Adjust the sampling frequency of distributed sensor nodes through adaptive traffic regulation algorithm; When the rainstorm warning level is lower than the preset warning level, the existing communication protocol and sampling frequency are maintained.

[0016] By implementing this technical solution, when the heavy rain warning level reaches or exceeds a preset threshold, the system automatically triggers a communication protocol switch command, switching the multi-source data transmission link to a high-priority communication protocol such as 5G slicing network and Beidou short message. High-priority communication protocols offer higher bandwidth, lower latency, and stronger interference resistance, ensuring the real-time transmission of critical data. When the heavy rain warning level is high, the high-priority communication protocol improves data transmission success rates and avoids warning delays or failures caused by communication interruptions. The low latency of the high-priority communication protocol shortens the transmission time of warning information, significantly improving the system's response speed. It also resists interference and ensures stable data transmission in extreme weather scenarios. After switching to the high-priority communication protocol, the system monitors the return quality of multi-source data in real time. If the return quality falls below the preset quality threshold, the system automatically deploys relay nodes to compensate for the transmission link, ensuring the integrity and stability of data transmission. This compensation mechanism of relay nodes improves the return quality compliance rate, reduces packet loss, and narrows the latency fluctuation range.

[0017] Furthermore, the relay node deployment methods include: Get candidate relay nodes; Detect coverage blind spots based on transmission links; Eliminate candidate relay nodes in coverage blind spots and anchor key relay nodes and backup relay nodes; When there are multiple key relay nodes, the key relay nodes at corresponding positions are activated according to the results of the multi-objective optimization algorithm.

[0018] By employing the aforementioned technical solution, the system leverages the environmental perception capabilities of a distributed sensor network to dynamically identify candidate relay nodes, such as drones, mobile base stations, and edge computing nodes. It then selects potential relay nodes based on real-time environmental information such as terrain, weather, and signal strength. The candidate node selection process is based on multi-source data fusion, including sensor data, Geographic Information System (GIS) data, and communication network topology data, ensuring a balanced distribution of candidate nodes and maximizing coverage. Dynamic candidate node selection enables the system to rapidly adjust relay node deployment based on scenarios such as terrain changes caused by heavy rain, signal interference, and environmental variations, adapting to complex scenarios. Multi-source data fusion ensures that the candidate node distribution maximizes coverage of transmission link blind spots. This dynamic selection mechanism avoids redundant deployment of fixed relay nodes, improving resource utilization. After eliminating candidate nodes in blind spots, the system identifies key and backup relay nodes based on metrics such as signal strength, location, and energy consumption. Key relay nodes ensure core coverage of the transmission link, while backup relay nodes provide redundancy in the event of key node failure.

[0019] In a second aspect, the present application provides a wireless network-based rainstorm warning system, which adopts the following technical solution: a wireless network-based rainstorm warning system includes an acquisition module for acquiring multi-source data and a historical database; A memory, used for storing a program of a control method of a rainstorm warning method based on a wireless network; The program in the memory can be loaded and executed by the processor to implement a control method for a rainstorm warning method based on a wireless network.

[0020] By implementing this technical solution, efficient collaboration between modules, memory, and processors is achieved, improving the scalability, data utilization, computing efficiency, and real-time performance of the rainstorm warning system. Compared with traditional methods, this system improves data fusion capabilities and the accuracy of graded warnings, extending warning lead times. Furthermore, the modular design and highly integrated architecture reduce system deployment and maintenance costs while improving resource utilization. This significantly enhances adaptability and cost-effectiveness in complex environments, providing technical support for urban waterlogging prevention, traffic scheduling, and personnel evacuation.

[0021] In a third aspect, the present application provides a smart terminal that adopts the following technical solution: An intelligent terminal comprises a memory and a processor. The memory stores a computer program that can be loaded by the processor and executed by a wireless network-based rainstorm early warning method.

[0022] In a fourth aspect, the present application provides a computer storage medium capable of storing a corresponding program, employing the following technical solution: A computer-readable storage medium stores a computer program that can be loaded by a processor and execute any one of the above-mentioned wireless network-based rainstorm warning methods.

[0023] In summary, this application includes at least one of the following beneficial technical effects: 1. A dynamic threshold function is generated through a spatiotemporal feature extraction model. This function is combined with historical databases and real-time data to dynamically adjust threshold parameters, avoiding the limitations of a single fixed threshold. Dynamic and static thresholds are weighted and integrated to output graded warning thresholds. This allows for timely triggering of warnings when short bursts of heavy rainfall have not reached the fixed threshold but have already caused waterlogging, reducing false alarm rates. Furthermore, in scenarios where long-term cumulative rainfall exceeds the threshold but the system fails to respond, risks can be identified in advance, reducing missed alarm rates. 2. Dynamically calculates fluctuations in sensor return data, corrects key deployment points, avoids blind spots or redundancy, optimizes sensor density and types based on geographic location and topographical features, and improves monitoring coverage. When transmission link quality degrades, it automatically deploys relay nodes and activates key relay nodes based on a multi-objective optimization algorithm to ensure the integrity and stability of data transmission, improve return quality compliance, and reduce packet loss. 3. Efficient collaboration among modules, memory, and processors, combined with edge computing and cloud computing architecture, enables efficient processing of multi-source data and real-time early warning, shortens response time, and modular design supports functional expansion and rapid maintenance, reduces deployment and operation and maintenance costs, and improves resource utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 This is a flowchart of a wireless network-based rainstorm warning method in an embodiment of the present application.

[0025] Figure 2 This is a flow chart of a method for deploying distributed sensor nodes in an embodiment of the present application.

[0026] Figure 3 4 is a flow chart of a method for updating a sampling frequency of a credible sensor in an embodiment of the present application.

[0027] Figure 4 This is a flow chart of a method for determining the validity of target data in an embodiment of the present application.

[0028] Figure 5 This is the process of the method for updating the graded warning threshold after outputting the target data in the embodiment of the present application. Figure 1 .

[0029] Figure 6 This is the process of the method for updating the graded warning threshold after outputting the target data in the embodiment of the present application. Figure 2 .

[0030] Figure 7This is a flowchart of the method for switching information transmission paths in an embodiment of the present application.

[0031] Figure 8 It is a flowchart of the deployment method of relay nodes in an embodiment of the present application.

[0032] Figure 9 It is a module diagram of a wireless network-based rainstorm warning method in an embodiment of the present application. DETAILED DESCRIPTION

[0033] In order to make the purpose, technical solutions and advantages of this application more clear, the following Figure 1-9 It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application.

[0034] The embodiment of the present application discloses a rainstorm warning method based on a wireless network. Figure 1 ,The rainstorm warning methods based on wireless networks include: Step 100: Collect multi-source data through distributed sensor nodes, where the multi-source data includes environmental data, drainage network flow data, and weather radar data.

[0035] Distributed sensor nodes refer to multiple sensor devices deployed according to certain rules within the monitoring area. These devices work together through network connections to collect and transmit multi-source data in real time to achieve comprehensive perception and dynamic monitoring of the target area.

[0036] Multi-source data refers to a collection of data with varying types and characteristics, originating from various sources. Data sources can include sensors, databases, historical records, and third-party platforms. Multi-source data has diverse sources, heterogeneous formats, and strong information complementarity. By integrating and analyzing multi-source data, a more comprehensive and accurate description of the target object or environmental state can be achieved.

[0037] In sensor networks, environmental sensors collect data such as temperature, humidity, air pressure, and light; water level sensors monitor water level changes in rivers, lakes, and drainage networks; rainfall sensors record meteorological data such as rainfall amount and intensity; displacement sensors detect geological changes such as landslides and building subsidence; and video surveillance uses image recognition technology to capture information such as water depth and traffic flow. Historical databases can be used to retrieve stored historical meteorological data such as rainfall, temperature, and wind speed; historical disaster records showing the time, location, and scale of floods and landslides; and geographic information such as terrain elevation, river distribution, and drainage network topology.

[0038] Real-time communication data can include real-time weather warnings from meteorological authorities, real-time traffic data such as congestion and vehicle flow, and real-time disaster reports from social media and news platforms. Third-party data platforms can access official data from meteorological bureaus, water conservancy bureaus, geological monitoring agencies, and other institutions, as well as satellite remote sensing data such as precipitation estimates and surface temperature, and terrain and road data from map service providers.

[0039] Multi-source data collection can be achieved through distributed sensor networks, enabling real-time acquisition of environmental, meteorological, and water level data. For example, rain gauges collect rainfall data every minute, while water level sensors monitor river level changes in real time. Historical meteorological, geological, and disaster data can be stored in databases as a foundation for analysis and modeling. For example, rainfall data from the past 10 years can be stored to analyze rainfall trends and extreme weather events. Real-time or historical data can be obtained from third-party platforms (such as meteorological bureaus and water conservancy bureaus). For example, access to real-time weather warning information from meteorological bureaus and satellite remote sensing data can be used. Data from these different sources can then be fused to generate higher-level information. For example, rainfall data can be combined with topographic data to generate a waterlogging depth prediction model. These diverse data sources provide multi-faceted information, avoiding the limitations of a single source. During heavy rainstorms, real-time rainfall data can dynamically adjust warning thresholds. Real-time data can dynamically reflect environmental changes, enabling timely adjustments to warning strategies.

[0040] Step 101: pre-process multi-source data to obtain real-time data.

[0041] Real-time data refers to data collected from multiple sources, such as distributed sensor networks, that, after preprocessing, instantly reflects current environmental conditions, such as rainfall, water level changes, and meteorological conditions. Multi-source data preprocessing is a fundamental step in building spatiotemporal feature extraction models. Through operations such as cleaning, synchronization, alignment, standardization, aggregation, and feature extraction, data quality and consistency are improved, providing high-quality data input for subsequent spatiotemporal feature modeling and analysis.

[0042] Step 102: associate and match the real-time data with the historical database to build a spatiotemporal feature extraction model.

[0043] The historical database is a collection of multi-source data such as meteorological, hydrological, and geographic information and historical disaster records of the target area in the past, which is used to analyze the laws of spatiotemporal evolution and provide a reference basis for current analysis.

[0044] Association matching refers to the process of aligning and associating real-time data with multi-source data in historical databases according to time, space, and feature dimensions to explore the spatiotemporal evolution patterns of data and build a spatiotemporal feature extraction model.

[0045] Spatiotemporal feature extraction models analyze and process multi-source spatiotemporal data containing information in both temporal and spatial dimensions, mining and extracting key features that exhibit temporal dynamics and spatial distribution characteristics. Spatiotemporal feature extraction models can analyze multi-source data such as environmental data, drainage network flow data, and weather radar data, mining the temporal and spatial characteristics of the data to support tasks such as rainstorm warnings and disaster prediction.

[0046] The construction method of the spatiotemporal feature extraction model includes five steps: data preparation, feature extraction, model selection and construction, training and optimization, and evaluation and application. By combining the dynamic changes in the temporal dimension with the distribution characteristics of the spatial dimension, and using traditional statistical methods or deep learning models (such as ST-CNN, LSTM, and ST-GNN), it can efficiently capture the spatiotemporal patterns of multi-source data, providing accurate support for heavy rain warnings and disaster predictions.

[0047] Step 103: Generate a dynamic threshold function based on the spatiotemporal feature extraction model, perform weighted fusion on the dynamic threshold function and the preset static threshold, and output a graded warning threshold.

[0048] The dynamic threshold function is an intelligent computing model that adjusts threshold parameters in real time to adapt to environmental changes, equipment status, and data characteristics based on multi-dimensional dynamic factors such as time, space, and environment of the data, including season, climate, geographic location, and sensor status. It is used to improve the adaptability of the monitoring system and the accuracy of early warnings.

[0049] Preset static thresholds are fixed, unchanging values ​​used to determine whether sensor data exceeds normal ranges (e.g., water levels exceeding warning levels or rainfall reaching warning levels). Static thresholds do not change over time, space, or environmental conditions; they trigger alerts based solely on pre-set parameters.

[0050] The method for weighted fusion of dynamic threshold function and static threshold is as follows: according to the real-time environmental adaptability reflected by the dynamic threshold function and the empirical benchmark represented by the static threshold, by dynamically adjusting the weight coefficient (such as based on data reliability assessment or scenario requirements), the dynamic threshold and the static threshold are linearly combined according to the weight to generate a comprehensive graded warning threshold, thereby achieving a balanced adaptation of empirical rules and real-time calculations.

[0051] The formula for double threshold weighted fusion is: T 融合 =α·T 动态 +(1-α)·T 静态 Among them, α is a dynamic weight, which is dynamically adjusted by the output of the spatiotemporal feature model or the scene index.

[0052] Step 104 , during the real-time monitoring process, determines whether the current water depth growth rate exceeds the gradient critical value of the graded warning threshold.

[0053] The growth rate of water depth refers to the change in water depth per unit time (minutes / hours), measured in cm / min or cm / h. It is used to measure the speed of water rise and is a key dynamic parameter in the rainstorm warning system for judging the disaster risk level and the urgency of emergency response.

[0054] The graded warning threshold is a weighted fusion of the dynamic threshold generated by the spatiotemporal feature extraction model and the preset static threshold. It is the critical threshold corresponding to different risk levels (such as yellow, orange, and red) divided by the growth rate of water depth. It is used to accurately judge the current waterlogging risk and trigger the corresponding level of emergency response.

[0055] The gradient critical value is the growth rate threshold in the graded warning threshold system, which is used to determine whether the growth rate of water depth has reached a specific risk level (such as upgrading from yellow warning to orange warning), reflecting the mutation limit of the water rise rate.

[0056] Step 105: If exceeded, the rainstorm warning level is output according to whether the current water depth growth rate exceeds the gradient critical value of the graded warning threshold.

[0057] By comparing the real-time growth rate with the classification critical value, the warning level with the highest matching level is dynamically matched to achieve accurate risk classification response.

[0058] For example, the gradient critical values ​​for the graded warning thresholds for a particular area are set as follows: Yellow Warning: Water Depth Increases ≥ 5cm / h; Orange Warning: Water Depth Increases ≥ 10cm / h; Red Warning: Water Depth Increases ≥ 20cm / h. Real-time monitoring shows that the current water depth increase is 15cm / h, exceeding the thresholds for yellow (5cm / h) and orange (10cm / h), but not reaching red (20cm / h). The system then triggers an orange warning, prompting relevant departments to initiate emergency drainage measures and notifying the public to take precautions.

[0059] Reference Figure 2 ,The deployment methods of distributed sensor nodes include: Step 200: Obtain key deployment points and deployment density based on the geographical location and topographic features of the monitoring area.

[0060] Key deployment points are core monitoring locations that are most sensitive to disasters such as rainstorm waterlogging, flash floods, and landslides, selected based on the monitoring area's geographic location (such as river confluences and low-lying, flood-prone areas), topography (such as valley exits and sudden slope changes), and historical disaster data. Examples include river bends (where vortexes and siltation are prone to form), drainage network intersections (where flow is concentrated), and historical waterlogging black spots.

[0061] Deployment density refers to the number of sensors installed per unit area and needs to be adjusted dynamically based on terrain complexity, historical disaster frequency, population / asset value density, and regional risk level. High-risk areas require high-density deployment, while low-risk areas require low-density deployment. For example, in mountainous areas prone to debris flow gullies and urban flooding, 5-10 sensors per square kilometer are recommended, while in plains and gentle slopes or historically disaster-free areas, 1-2 sensors per square kilometer are recommended.

[0062] Key points ensure monitoring coverage of the core disaster area, while deployment density guarantees data accuracy and timely response.

[0063] Step 201: Determine sensor categories based on key deployment points.

[0064] Sensor category refers to the type of sensor with specific monitoring capabilities selected based on the key deployment points in the monitoring area (such as terrain, function, and risk level), which is used to accurately collect corresponding disaster elements.

[0065] Sensor categories include radar water level gauges, pressure water level gauges and ultrasonic water level gauges for monitoring water levels, tipping bucket rain gauges and optical rain gauges for monitoring rainfall, soil moisture sensors, tilt sensors and infrasound sensors for monitoring soil moisture and geological conditions, video surveillance cameras and AI analysis equipment and drone inspection systems for auxiliary verification, as well as multi-parameter integrated environmental monitoring stations and IoT gateways for comprehensive monitoring. These sensor categories are specifically configured according to the key deployment points and risk types in the monitoring area to achieve accurate monitoring of various disasters such as rainstorm waterlogging, mountain torrents, and landslides.

[0066] In step 202 , after sensors are deployed in the monitoring area according to key deployment points, deployment density, and sensor types, actual data fluctuation values ​​of the returned data are calculated based on the returned data of corresponding sensors deployed in the monitoring area.

[0067] Backhaul data refers to the raw monitoring data collected in real time by various sensors (such as water level gauges and rain gauges) deployed in the monitoring area and transmitted to the monitoring center via wireless networks. It includes values ​​such as water level height, rainfall, and soil moisture, and is the basic data source for subsequent calculations and analysis.

[0068] The actual data fluctuation value refers to a quantitative indicator reflecting the dynamic changes in data, calculated after preprocessing sensor data, such as denoising and removing outliers. Fluctuations in the temporal dimension are calculated by comparing the difference between data at adjacent time points (e.g., current water level minus the water level at the previous moment) or by using statistical standard deviations. Fluctuations in the spatial dimension are calculated by analyzing the differences in data from multiple sensors within the same area (e.g., the difference in water levels at adjacent points). Comprehensive fluctuations are calculated by combining historical baseline data and calculating the magnitude of the current data's deviation from the long-term mean (e.g., current rainfall / historical mean for the same period). The calculation of actual data fluctuation values ​​can identify abnormal sudden changes, such as short-term water level surges, as well as trend changes, such as continuous rainfall accumulation, providing a basis for dynamic threshold adjustment and risk level determination. For example, when the water level fluctuation value exceeds the preset threshold, an alert upgrade is triggered. When the rainfall fluctuation value remains persistently high, the monitoring frequency is increased.

[0069] Step 203: determine whether the actual data fluctuation value is greater than a preset fluctuation threshold.

[0070] The dynamic fluctuation amplitude (such as short-term water level increase and rainfall growth rate) calculated from the real-time sensor data after preprocessing (such as denoising and outlier removal) is compared with the system's preset fluctuation threshold. If the actual fluctuation value exceeds the threshold, an early warning response is triggered (such as upgrading the warning level and increasing the monitoring frequency). If the actual fluctuation value is less than or equal to the threshold, the current monitoring status is maintained. By dynamically comparing real-time fluctuations with preset thresholds, abnormal risks can be quickly identified, improving the timeliness and accuracy of disaster warnings.

[0071] Step 204: If yes, correct the key deployment points.

[0072] Step 205: If the answer is no, there is no need to modify the key deployment points.

[0073] When the actual fluctuation value of the data sent back by the sensor exceeds the system's preset fluctuation threshold (such as a short-term surge in water level or a surge in rainfall), the system automatically analyzes the monitoring area corresponding to the abnormal data and dynamically adjusts key deployment points.

[0074] Methods for correcting key deployment points include: intensifying monitoring of high-risk areas, such as adding sensors around points with abnormal fluctuations to capture details of local mutations (increased from 1 per kilometer to 3); optimizing point layout, for example, if fluctuations originate from terrain blind spots (such as river bends that are not covered), adding new sensors to fill monitoring gaps; replacing failed equipment, for example, if sensor failures are caused by fluctuations and abnormal data appear (such as data mutations that have no physical meaning), triggering the replacement and calibration of backup equipment.

[0075] For example, the water level fluctuation in a low-lying area reaches 1.2m / hour (threshold 0.8m / hour). The system adds a radar water level meter next to the origin and can also link with a drone to verify the terrain to ensure data accuracy.

[0076] Reference Figure 3 , the rainstorm warning method based on wireless network also includes: Step 300: Obtain sensor data and historical climate information corresponding to the monitoring area from a historical database.

[0077] The historical database is a professional data system that stores the raw data collected by sensors in the monitoring area over a long period of time (such as water level, rainfall, soil moisture, etc.) and the corresponding historical climate information (such as rainfall patterns and temperature changes). It has time series characteristics and spatial correlation.

[0078] Historical climate information refers to the long-term record of weather conditions in a certain area in the past, including statistical data of meteorological elements such as temperature, precipitation, wind speed, and archives of extreme climate events. It is used to analyze regional climate patterns and provide a reference for disaster warning. By comparing current monitoring values ​​with historical data, abnormal fluctuation trends can be identified, assisting in dynamic threshold adjustment and disaster warning classification.

[0079] Step 301: Establish a correlation function between sensor data and historical climate information.

[0080] A correlation function is established between sensor data (such as water level, rainfall, and soil moisture) and historical climate information (such as rainfall, temperature, and extreme weather events) to quantify the statistical association between the two. The formula is as follows: C S,C =f(S,C)=α·Pearson(S,C)+β·Spearman(S,C)+γ·time-lagged correlation(S,C) Where S = sensor data (e.g., water level, rainfall); C = historical climate information (e.g., rainfall, temperature); Pearson (S, C) = Pearson correlation coefficient (linear correlation); Spearman (S, C) = Spearman rank correlation coefficient (nonlinear correlation); time lag correlation (S, C) = taking into account the lag effect of climate events on sensor data (e.g., water level rise 1-3 days after rainfall); α, β, γ = weight coefficients (which can be optimized based on historical data fitting).

[0081] Pearson correlation coefficient (linear relationship): Spearman correlation coefficient (non-linear relationship): calculated based on data rank, suitable for non-linear but monotonic relationships; Time lag analysis (rainfall and water level delay effect): Calculate the effect of C(t) on S(t+Δt) (Δt=1, 2, 3… days).

[0082] The overall relevance score is as follows: time-lagged correlation (S,C); Among them, the weights α, β, and γ are optimized by fitting historical data.

[0083] For example, predicting the relationship between a river's water level and rainfall: Input data: Sensor data S: water level of a river section (meters); Historical climate information C: Daily rainfall in the past five years (mm) Calculation results: Pearson(S,C)=0.85(strong linear correlation) 1. Time lag correlation: The water level rises most significantly on the second day after rainfall (Δt = 2) 2. Comprehensive relevance score: 3.C S,C =0.5·0.85+0.3·0.88+0.2·0.92=0.87 (highly correlated) 4. When real-time monitoring shows that rainfall exceeds the historical threshold, the water level rise trend is predicted in combination with the correlation function, and the warning threshold is dynamically adjusted.

[0084] Step 302: Obtain corresponding weather warning information.

[0085] Weather warning information refers to disastrous weather alerts issued by meteorological departments based on real-time monitoring data and forecasting models. It is used to remind the public and relevant departments to guard against possible extreme weather events (such as heavy rain, typhoons, high temperatures, etc.). Weather warning information is one of the important inputs of the dynamic threshold model. The system can adjust the sensor monitoring frequency and warning response strategy based on the warning level.

[0086] For example, when an orange rainstorm warning is received, the water level monitoring frequency is increased to 1 minute / time, and the emergency drainage plan is activated; when a red high temperature warning is received, forest fire risk monitoring is strengthened, and the fire department is on standby.

[0087] Step 303: anchor the trusted sensor based on the weather warning information and the correlation function.

[0088] Trusted sensors are sensors that have high data reliability and strong correlation with warning events in specific disaster scenarios, based on a historical correlation function between weather warning information and sensor data. The data from these sensors serves as the core basis for dynamic threshold adjustment and graded warnings.

[0089] The method for anchoring trusted sensors is as follows: First, extract the key parameters of the weather warning (such as the magnitude of the rainstorm, the scope of impact, and the duration), and match them with the sensor data features corresponding to similar warning events in the historical database (such as the rate of water level surge and peak rainfall). Secondly, use the pre-built correlation function to calculate the correlation score of each sensor under the current warning event, and give priority to sensors with high correlation (for example, in the case of a rainstorm warning, the correlation between rain gauge and weather radar data is given priority). Then, eliminate outliers that are highly inconsistent with the data of neighboring sensors (such as a sudden rise in water level at a certain point but no change in the surrounding area), and use the geographic information system (GIS) to verify whether the sensor location is within the warning core area. Finally, give higher weights to highly correlated sensors (for example, the weight is increased to 80% under a red warning), and reduce the weight or block low-correlation sensors.

[0090] For example, if an orange rainstorm warning (24-hour rainfall ≥ 100 mm) is issued for a region, affecting Area A, where rainfall intensity is high, the system prioritizes rain gauge and weather radar data within Area A (correlation score 0.9), while eliminating soil moisture sensors (score 0.3) located far from the warning area. This also excludes abnormal spikes in data from a rain gauge due to equipment failure. Ultimately, five highly reliable sensors are anchored for dynamic threshold calculation and graded warnings.

[0091] Step 304: Send a sampling frequency update instruction to the trusted sensor. The sampling frequency update instruction is used to instruct the trusted sensor to increase the sampling frequency.

[0092] After the system anchors trusted sensors through weather warning information and correlation functions, it sends instructions to these sensors to dynamically adjust the sampling frequency to improve the real-time performance and monitoring accuracy of key data. The instructions contain at least the following core parameters: target sensor ID, new sampling frequency, effective time and duration. The instruction example is: { "command":"UPDATE_SAMPLING_RATE", "sensor_id":"WS-001", "new_rate":"10 times / minute", "start_time":"20XX-XX-XXT14:00:00Z", "duration":"2 hours", "reason":"Orange rainstorm warning response" } The 10-minute sampling frequency captures sub-second changes in flood processes, while the 2-hour duration covers the duration of a typical rainstorm event. Clear start and end times facilitate automated system scheduling, and the "Orange Rainstorm Warning Response" designation facilitates auditability. This command is ideal for disaster response phases requiring rapid response, significantly improving monitoring accuracy during critical periods without increasing long-term operating costs.

[0093] Reference Figure 4 , the rainstorm warning method based on wireless network also includes: Step 400: Acquire historical location information and current location information of the target sensor.

[0094] Target sensors refer to sensor devices that are monitored in the disaster warning system. Their physical locations may change (be washed away or displaced) due to extreme weather (such as floods and mudslides). Their status needs to be determined by comparing historical and current locations.

[0095] Historical location information refers to the initial coordinates and historical trajectory data when the sensor was deployed. Current location information refers to real-time monitoring or the most recently acquired sensor location data.

[0096] In step 401 , when the displacement between the historical position information and the current position information is greater than a preset displacement threshold, sensors near the target sensor are defined as neighboring sensors.

[0097] The preset displacement threshold is a critical value set in the sensor displacement monitoring system to determine whether a target sensor has experienced significant displacement (such as being washed away or experiencing a landslide). When the displacement between the target sensor's historical position and its current position exceeds this threshold, the adjacent sensor linkage mechanism is triggered to assist in risk assessment and correct monitoring data.

[0098] The preset displacement threshold represents the maximum safe displacement that the sensor can tolerate. For example, in landslide monitoring, the threshold may be set to 10 centimeters, and in flood erosion scenarios it may be set to 1 meter. If the threshold is exceeded, it is considered that the sensor has failed or the environment has changed significantly.

[0099] Step 402 : Acquire a target data value of a target sensor and adjacent data values ​​of adjacent sensors.

[0100] Target data values ​​refer to the key environmental parameters currently being collected by the target sensor being monitored, such as the current water level of a river water level gauge (ID: WS-001) of 3.2 meters. Neighboring data values ​​refer to data collected by similar sensors surrounding the target sensor (defined by spatial proximity) and are used for cross-validation and data compensation. For example, neighboring sensors for target sensor WS-001 include WS-002 (water level: 3.15 meters), WS-003 (water level: 3.22 meters), and WS-004 (water level: 3.18 meters).

[0101] In the event of a sensor failure, historical data from neighboring sensors is extracted for compensation calculations. Cross-validation of multi-source data improves monitoring reliability in extreme environments. If the target water level (3.2 meters) deviates from the mean value of neighboring sensors (3.18 meters) by more than a threshold (e.g., 0.1 meters), an anomaly is flagged. If the target sensor is washed away, the time series is reconstructed using interpolated data from neighboring sensors. A sudden, simultaneous rise of neighboring sensors (e.g., 0.5 meters / minute) triggers a high-level alert.

[0102] Step 403: Determine whether the target data value is valid based on the neighboring data values.

[0103] By comparing the consistency of the target sensor data with that of neighboring sensors, we determine whether the target data has become invalid due to equipment failure, environmental interference, or unexpected events (such as a sensor being washed away). Validity is determined using a spatial weighted average method, which calculates the mean of neighboring data weighted by distance. If the target value deviates significantly from the weighted mean, the data is considered invalid.

[0104] Step 404: If yes, output the target data.

[0105] If the target data is valid after verification by neighboring data, the target data is output and the original value is retained for subsequent analysis.

[0106] Step 405: If no, the target data is eliminated.

[0107] If the target data fails to pass the neighboring data verification and is invalid, the target data will be eliminated, marked as abnormal, and will not participate in the calculation, and the compensation mechanism will be triggered.

[0108] Reference Figure 5 and Figure 6 The rainstorm warning method based on wireless network also includes, after outputting the target data: Step 500: Update the spatiotemporal feature extraction model based on the target data and the neighboring data values.

[0109] When the target data and the adjacent data are verified, that is, the target data is valid, the system automatically triggers the update process of the spatiotemporal feature extraction model to ensure that the model parameters adapt to the latest monitoring environment.

[0110] For example, the initial model for water level monitoring in a particular river basin used data from five adjacent sensors. During a rainstorm, three temporary sensors were added. The system automatically expanded the spatial range to 800 meters, added a mutation detection feature, and adjusted the time window to 15 minutes. This enabled the spatiotemporal feature extraction model to maintain high-precision prediction capabilities even in extreme weather conditions, providing a reliable foundation for graded early warnings.

[0111] Step 501: Generate a dynamic threshold update function based on the updated spatiotemporal feature extraction model.

[0112] Based on the updated spatiotemporal feature extraction model, the warning threshold under the current environment is dynamically calculated to ensure that the threshold can reflect the actual risk status of the monitored area in real time.

[0113] Function input parameters include the basic static threshold T base , time dimension feature F temporal , spatial dimension feature F spatial , time feature weight W temporal , spatial feature weight W spatial , environmental correction coefficient α. Among them, T base Derived from preset configuration, such as historical experience value; F temporal Derived from spatiotemporal feature models, such as rate of change, trend; F spatial Derived from spatiotemporal feature models, such as neighboring sensor differences and gradients; W temporal and W spatial is a dynamically adjusted value; α comes from real-time environmental data, such as rainfall intensity and terrain impact.

[0114] The calculation formula of dynamic threshold is: T dynamic =T base ×(1+W temporal ·F temporal +W spalial ·F spatial +α) Time dimension adjustment F temporal Calculation is performed based on the time rate of change of target data (such as the rate of water level rise), for example If the current water level rises twice as fast as the historical rate, then F temporal =1; Spatial dimension adjustment F spatia base l The calculation is based on the dispersion of neighboring sensor data (such as standard deviation), for example If the dispersion is high, increase the threshold; The environmental correction factor α needs to take into account extreme weather (such as red alert for rainstorms) or terrain effects (such as steep slopes), for example Application example: when monitoring the water level of a river, the basic threshold T base =3.0m (historical warning water level), time characteristic F temporal = 0.8 (the current water level rise rate is 1.8 times that of the history), spatial feature F spatial = 0.3 (high dispersion of adjacent sensor data), environmental correction α = 0.2 (red alert): The calculation is as follows, T dynamic =3.0×(1+0.6×0.8+0.4×0.3+0.2)=4.68m, the output dynamic threshold is adjusted to 4.68 meters, and the system triggers an orange warning.

[0115] Step 502: Perform weighted fusion on the output value of the dynamic threshold update function and the graded warning threshold, and output the updated graded warning threshold.

[0116] Step 5021: Update the output value of the dynamic threshold function (T dynamic ) and the preset graded warning threshold (T static ) to perform weighted fusion and generate a more accurate updated graded warning threshold (T final ), taking into account both real-time and empirical rules.

[0117] The fusion formula is, T final =ω dynamic ·T dynamic +ω static ·T static Among them, ω dynamic is the dynamic threshold weight (0≤ω dynamic ≤1); cita ω ts is the static threshold weight (ω static =1-ω dynamic ).

[0118] Step 5022: The static threshold weight ω is adjusted based on the data reliability, the severity of environmental changes, and historical verification results. static and dynamic threshold weight ω dynamic Make dynamic adjustments.

[0119] The weight calculation method for adjusting the data reliability is as follows: based on the sensor data quality such as noise level and outlier ratio, for example, if the target sensor data is highly reliable, ω dynamic →1; The weight calculation method for adjusting according to the severity of environmental changes is as follows: adjustment is made based on the intensity of rainstorms and sudden changes in terrain. For example, during heavy rainfall, ω dynamic →0.8, when ω is stable dynamic →0.5; The weight calculation method for adjusting based on historical verification results is: dynamic optimization based on historical warning accuracy. For example, if the dynamic threshold warning is more accurate, then gradually increase ω dynamic .

[0120] Step 5023: merge and output the updated graded warning thresholds.

[0121] For example, in a certain urban flood monitoring system, the dynamic threshold is calculated by the spatiotemporal feature model as T dynamic =4.68m, the static threshold is obtained from historical experience T static =4.0m, based on the current environment ω dynamic =0.7 is confirmed as an orange rainstorm warning. The calculation process of the graded warning threshold is as follows: T final =0.7×4.68+0.3×4.0=4.476m The updated graded warning threshold is rounded to 4.48 meters.

[0122] Reference Figure 7 , the method for switching the information transmission path includes: Step 600 , when the rainstorm warning level is greater than or equal to the preset warning level, triggering a communication protocol switching instruction to control the transmission link of the multi-source data to switch to a high-priority communication protocol.

[0123] The rain warning level is a measure of the severity of rainstorms issued by a meteorological monitoring system or warning platform. It is typically categorized into different levels (e.g., blue, yellow, orange, and red), with higher values ​​indicating greater intensity. The preset warning level is a threshold set by the system to determine whether to switch communication protocols. For example, if the preset value = Orange Warning (Level 3), a switch is triggered when the actual warning level ≥ Orange.

[0124] Communication protocol switching commands are system-generated instructions that control the transmission link of multi-source data (such as sensors, cameras, and radar) from the current protocol to a higher-priority protocol. These commands can be issued via an edge computing gateway or cloud platform to ensure the reliable transmission of critical data (such as water level and rainfall) even in harsh environments.

[0125] High-priority communication protocols are designed for high reliability and typically feature low latency, high bandwidth, and anti-interference capabilities. For example, in extreme conditions like heavy rain, the new air interface protocol switches from 4G to 5G NR to ensure data integrity.

[0126] Step 601 , after the transmission link is switched to a high priority communication protocol, the quality of the return transmission of multi-source data is detected.

[0127] Backhaul quality refers to the integrity and reliability of data transmitted from sensors / devices to monitoring centers or cloud platforms. It is comprehensively evaluated using metrics such as packet loss rate, latency, bandwidth utilization, error rate, and signal strength. If backhaul quality falls below a threshold (e.g., a score <70%), critical data may not reach the monitoring center in a timely manner, leading to misjudgments due to missing or erroneous data.

[0128] The comprehensive score is calculated using weighted arithmetic. For example, quality score = 0.4 × (1-packet loss rate) + 0.3 × (1-delay / 100) + 0.3 × (1-error rate). A score < 70% is considered low quality.

[0129] Step 602: When the backhaul quality is lower than a preset quality threshold, deploy a relay node for compensating the transmission link.

[0130] The preset quality threshold is a system-defined standard for data backhaul quality, used to determine whether the current transmission link meets reliability requirements. When actual backhaul quality indicators (such as packet loss rate, latency, and error rate) exceed this threshold, a compensation mechanism (deployment of relay nodes) is triggered. Relay nodes are auxiliary devices deployed along the transmission path to enhance signal coverage, optimize routing, or cache data, ensuring reliable backhaul of multi-source data in harsh environments.

[0131] When quality is detected to be substandard, relay node deployment is triggered. Once the relay nodes are in place, the backhaul quality is reassessed. If it still falls short, the number of nodes is increased or the protocol is switched. Once quality returns to normal, relay node resources are automatically recycled to reduce system energy consumption.

[0132] Step 603: Adjust the sampling frequency of the distributed sensor nodes through an adaptive traffic regulation algorithm.

[0133] The adaptive traffic regulation algorithm is an intelligent mechanism that dynamically adjusts the amount of data transmission. It automatically optimizes the data sending strategy of multi-source sensor nodes based on network conditions, device load and data importance to ensure that critical data is transmitted first while reducing the risk of network congestion.

[0134] For example, during heavy rain, key data such as water level and rainfall are prioritized for transmission, reducing the frame rate of video surveillance. When the network is congested, data packets are automatically compressed or redundant sampling points are skipped.

[0135] The sampling frequency refers to the rate at which a sensor collects data (unit: Hz or times / minute), that is, the number of times data is acquired per unit time. Dynamically adjusting the sampling frequency can ensure data validity while reducing redundant transmissions.

[0136] The adaptive traffic regulation algorithm dynamically adjusts the sampling frequency based on network conditions and data priority. Critical data (such as water levels) is continuously transmitted at a high frequency to ensure timely warnings, while non-critical data (such as device status) is transmitted at a low frequency to conserve bandwidth resources.

[0137] Step 604 : When the rainstorm warning level is lower than the preset warning level, maintain the existing communication protocol and sampling frequency.

[0138] When the rainstorm warning level is lower than the preset warning level (for example, the current warning is yellow, and the preset threshold is orange or above), the system determines that the current environmental risk is low and there is no need to switch to a high-priority communication protocol or adjust the sampling frequency, so the existing configuration remains unchanged.

[0139] Reference Figure 8 , the deployment methods of relay nodes include: Step 700: Acquire candidate relay nodes.

[0140] Candidate relay nodes are potential relay devices that the system initially screens after detecting coverage blind spots on a transmission link and can be used to enhance signal coverage and optimize data transmission paths. Candidate relay nodes can be fixed infrastructure, mobile devices, or temporarily deployed devices. Fixed infrastructure includes pre-deployed LoRa gateways, 5G micro base stations, and Wi-Fi repeaters; mobile devices include drones, emergency communication vehicles, and transportable LoRa nodes; and temporarily deployed devices include portable satellite communication terminals and temporarily installed signal boosters.

[0141] The acquisition method is to retrieve from the system's preset relay node library (such as the fixed gateway location in the city's emergency communication network), discover surrounding available devices through wireless signal scanning (such as Wi-Fi probes, Bluetooth beacons), obtain available shared relay resources from the remote management platform (such as 5G temporary base stations opened by operators), and receive relay node information manually deployed by on-site personnel (such as equipment temporarily set up during emergency response).

[0142] Each candidate relay node must contain the following information for subsequent screening: location coordinates (latitude and longitude / three-dimensional positioning), signal strength (transmission power, coverage range), bandwidth capability (maximum transmission rate), power consumption status (whether it supports long-term operation), and protocol compatibility (supported communication protocol types).

[0143] For example, when heavy rain causes signal interruption in mountainous areas, the system may obtain the following candidate relay nodes: three preset LoRa gateways at locations A, B, and C, two nearby emergency communication vehicles at locations D and E, and a hovering drone at location F.

[0144] Step 701: Detect coverage blind spots based on transmission links.

[0145] A coverage blind spot refers to an area where the communication link is interrupted or the quality is severely degraded due to physical obstacles, signal attenuation, or equipment failure during data transmission.

[0146] Step 702: Eliminate candidate relay nodes in the coverage blind area and anchor key relay nodes and backup relay nodes.

[0147] Key relay nodes are core nodes that cover the edges or interiors of blind spots and directly enhance signal coverage in the target area. They ensure stable data transmission back to the primary network within the blind spot and are therefore prioritized for deployment. Backup relay nodes are redundant nodes located near key relay nodes, such as mobile devices and drones, that can temporarily take over communications. They provide backup in the event of key node failures, enhancing system robustness.

[0148] The anchoring method is to exclude candidate nodes located in the center of the blind spot with insufficient signal strength and filter out nodes that cannot cover the blind spot due to occlusion. The optimal node is selected based on a comprehensive score of coverage range, signal strength, and bandwidth capacity to ensure that the candidate node can cover at least 80% of the blind spot. Redundant nodes are deployed around key nodes, and mobile devices such as drones are given priority to enhance dynamic adaptability.

[0149] For example, the radius of blind spot A is 200 meters. The LoRa gateway at position X, which covers 90% of the blind spot, is anchored as a key node. The drone at position Y, 80 meters away from X, can quickly respond to failures and is anchored as a backup node.

[0150] Step 703 : When there are multiple key relay nodes, activate the key relay nodes at corresponding locations according to the result of the multi-objective optimization algorithm.

[0151] When there are multiple key relay node candidates, the system needs to comprehensively consider ensuring the widest signal coverage in the blind spot, extending the relay node life, reducing maintenance costs, giving priority to low-cost equipment, and ensuring real-time data transmission, so as to select the optimal deployment plan.

[0152] The algorithm is a weighted multi-objective optimization algorithm. The input parameters include the location, signal strength, bandwidth, power consumption, etc. of each candidate relay node, as well as the blind spot coverage requirements such as the target area range and priority. The optimization objective function is: MaximizeF=ω1·C+ω2·(1 / E)+ω3·(1+C ost )+ω4·(1 / D), where C is the coverage (blind spot coverage rate), E is the energy consumption (power consumption per unit time), and C ost is the deployment cost, D is the average delay, ω1, ω2, ω3, and ω4 are all weight coefficients. Each weight coefficient needs to be dynamically adjusted according to the scenario. For example, ω4 is higher in the emergency scenario.

[0153] When the total number of nodes ≤ budget limit and the coverage range of a single node ≤ maximum transmission distance, output the set of activated key relay nodes (location, number).

[0154] The activation method begins by screening candidate nodes. Based on the results of a multi-objective optimization algorithm, these nodes are prioritized, prioritizing those with low energy consumption and covering the core of the blind spot. Activation is then carried out in stages. In the first stage, key nodes covering the center of the blind spot are activated to ensure connectivity in the core area. In the second stage, backup nodes at the edge of the network are activated as needed to fill coverage gaps. Finally, network quality is monitored in real time. If a node is overloaded, adjacent backup nodes are activated to share traffic.

[0155] For example, blind spot A needs to cover an area with a radius of 200 meters, and there are three candidate nodes: node X (located at the center, coverage 90%, high power consumption); node Y (located at the edge, coverage 60%, low power consumption); node Z (mobile drone, coverage 80%, high cost).

[0156] The optimization result is: activated node X + node Y (weight ω1 = 0.5, ω2 = 0.3, ω4 = 0.2). The activation result is used to actually deploy relay nodes and is fed back to step 700 to update candidate relay nodes.

[0157] Based on the same inventive concept, an embodiment of the present application provides a rainstorm warning system based on a wireless network.

[0158] Reference Figure 9 ,A wireless network based rainstorm warning system includes : an acquisition module for acquiring multi-source data and a historical database; A memory, used for storing a program of a control method of a rainstorm warning method based on a wireless network; The program in the memory can be loaded and executed by the processor to implement a control method for a rainstorm warning method based on a wireless network.

[0159] Those skilled in the art will clearly understand that for the sake of convenience and brevity, the division of the above-mentioned functional modules is only used as an example for illustration. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working processes of the above-mentioned systems, devices, and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0160] An embodiment of the present application provides a computer-readable storage medium storing a computer program that can be loaded by a processor and executed by a wireless network-based rainstorm warning method.

[0161] Computer storage media include, for example, various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0162] Based on the same inventive concept, an embodiment of the present application provides a smart terminal, including a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and execute a wireless network-based rainstorm warning method.

[0163] Those skilled in the art will clearly understand that for the sake of convenience and brevity, the division of the above-mentioned functional modules is only used as an example for illustration. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working processes of the above-mentioned systems, devices, and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0164] The above are all preferred embodiments of the present application and are not intended to limit the scope of protection of this application. Unless otherwise stated, any feature disclosed in this specification (including the abstract and drawings) may be replaced by other equivalent or similar features. In other words, unless otherwise stated, each feature is merely an example of a series of equivalent or similar features.

Claims

1. A rainstorm warning method based on wireless network, characterized in that: include: Collect multi-source data through distributed sensor nodes, including environmental data, drainage network flow data, and weather radar data; Preprocess multi-source data to obtain real-time data; Correlate and match real-time data with historical databases to build a spatiotemporal feature extraction model; Generate a dynamic threshold function based on the spatiotemporal feature extraction model, and perform weighted fusion of the dynamic threshold function with the preset static threshold to output a graded warning threshold; During real-time monitoring, determine whether the current water depth growth rate exceeds the gradient critical value of the graded warning threshold; If it exceeds, the heavy rain warning level will be output based on whether the current water depth growth rate exceeds the gradient critical value of the graded warning threshold.

2. The method for rainstorm warning based on wireless network according to claim 1, characterized in that: Distributed sensor node deployment methods include: Obtain key deployment points and deployment density based on the geographical location and topographic features of the monitoring area; Determine sensor categories based on key deployment points; After deploying sensors in the monitoring area according to key deployment points, deployment density, and sensor types, calculate the actual data fluctuation value of the returned data based on the return data of the corresponding sensors deployed in the monitoring area; Determine whether the actual data fluctuation value is greater than the preset fluctuation threshold; If yes, correct the key deployment points.

3. The method for rainstorm warning based on wireless network according to claim 2, characterized in that: The method further comprises: Obtain sensor data and historical climate information corresponding to the monitoring area in the historical database; Establish a correlation function between sensor data and historical climate information; Obtain corresponding weather warning information; Anchor trusted sensors based on weather warning information and correlation functions; A sampling frequency update instruction is sent to the trusted sensor, where the sampling frequency update instruction is used to instruct the trusted sensor to increase the sampling frequency.

4. The method for rainstorm warning based on wireless network according to claim 3, characterized in that: The method further comprises: Obtain historical location information and current location information of the target sensor; When the displacement between the historical position information and the current position information is greater than a preset displacement threshold, defining sensors near the target sensor as neighboring sensors; Obtaining a target data value of a target sensor and adjacent data values ​​of adjacent sensors; Determine whether the target data value is valid based on the adjacent data values; If yes, output the target data; If not, the target data is eliminated.

5. The method for rainstorm warning based on wireless network according to claim 4, characterized in that: After outputting the target data, the method further includes: Update the spatiotemporal feature extraction model based on the target data and neighboring data values; Generate a dynamic threshold update function based on the updated spatiotemporal feature extraction model; The output value of the dynamic threshold update function is weightedly fused with the graded warning threshold to output the updated graded warning threshold.

6. The method for rainstorm warning based on wireless network according to claim 1, characterized in that: The method of switching the information transmission path includes: When the rainstorm warning level is greater than or equal to the preset warning level, the communication protocol switching instruction is triggered to control the transmission link of multi-source data to switch to the high-priority communication protocol; After the transmission link switches to a high-priority communication protocol, the quality of multi-source data return is tested; When the backhaul quality falls below a preset quality threshold, a relay node is deployed to compensate for the transmission link; Adjust the sampling frequency of distributed sensor nodes through adaptive traffic regulation algorithm; When the rainstorm warning level is lower than the preset warning level, the existing communication protocol and sampling frequency are maintained.

7. The method for rainstorm warning based on wireless network according to claim 6, characterized in that: Relay node deployment methods include: Get candidate relay nodes; Detect coverage blind spots based on transmission links; Eliminate candidate relay nodes in coverage blind spots and anchor key relay nodes and backup relay nodes; When there are multiple key relay nodes, the key relay nodes at corresponding positions are activated according to the results of the multi-objective optimization algorithm.

8. A rainstorm warning system based on wireless network, characterized in that: include: Acquisition module, used to obtain multi-source data and historical database; A memory for storing a program of a control method for a wireless network-based rainstorm warning method according to any one of claims 1 to 7; The program in the memory can be loaded and executed by the processor to implement the control method of the wireless network-based rainstorm warning method as claimed in any one of claims 1 to 7.

9. An intelligent terminal, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executes any one of the wireless network-based rainstorm warning methods according to claims 1 to 7.

10. A computer-readable storage medium, characterized in that The device stores a computer program that can be loaded by a processor and executes the wireless network-based rainstorm warning method according to any one of claims 1 to 7.

Images