Edge internet-of-things rural water body total phosphorus content monitoring method and system

Abstract

The application relates to the technical field of phosphorus content monitoring, and discloses an edge internet-of-things rural water body total phosphorus content monitoring method and system.The method comprises the following steps: collecting sensing data of rural water bodies at edge node positions; using a lightweight deep inference model to perform feature extraction and total phosphorus content evaluation on the sensing data of the rural water bodies, and constructing transmission data packets by the edge nodes; based on a cooperative protocol and a time slot allocation algorithm between the edge nodes, the transmission strategy of the transmission data packets is dynamically adjusted; the historical transmission data packets of the edge nodes are fused, the total phosphorus content in the transmission data packets is checked and adaptively corrected by using an edge monitoring correction mode, and the corrected total phosphorus content is output.The application combines adaptive correction, lightweight deep inference, transmission strategy optimization and space-time checking mechanisms, realizes real-time accurate monitoring and automatic correction of abnormal total phosphorus content, and effectively improves the reliability, stability and edge computing efficiency of the monitoring data.

Description

Technical Field

[0001] This invention relates to intelligent sensing systems, particularly to the field of phosphorus content monitoring, and specifically to a method and system for monitoring total phosphorus content in rural water bodies using edge IoT. Background Technology

[0002] With the continuous increase in domestic sewage, agricultural non-point source pollution, and aquaculture discharge in rural areas, the problem of phosphorus enrichment in rural rivers, ditches, and small water bodies has become increasingly prominent. Total phosphorus content has become an important indicator for evaluating the eutrophication level and ecological security of rural water bodies. Traditional rural water quality monitoring mainly relies on manual sampling and laboratory chemical analysis methods. Although these methods offer high measurement accuracy, they suffer from problems such as long sampling cycles, limited monitoring range, lack of direct phosphorus content measurement equipment, and difficulty in achieving continuous online monitoring. These methods fail to meet the needs of long-term, real-time, and large-scale monitoring of dispersed rural water bodies. With the development of Internet of Things (IoT) technology, online water quality monitoring based on edge nodes and multi-source sensors has gradually become a research hotspot. By deploying sensing devices at key locations in rivers, real-time acquisition and remote transmission of water parameters can be achieved, providing data support for rural water environment management.

[0003] Existing sensor-based water quality monitoring systems still face several technical challenges in practical applications. For example, edge nodes for rural water monitoring are usually deployed in complex outdoor environments, and sensors are susceptible to factors such as temperature, humidity, and long-term operational drift, which can lead to deviations or drift in the collected data, affecting the reliability of the monitoring results. Furthermore, edge nodes typically rely on low-power wireless networks for data transmission. Under conditions of limited bandwidth and node power consumption, how to achieve efficient collaborative transmission and scheduling of data from multiple nodes remains a key issue.

[0004] Existing research, such as patent CN115358695A, proposes a monitoring technology for rural non-point source pollution (NPG) water bodies based on hyperspectral remote sensing. This technology determines the study area through information surveys and data collection, acquires hyperspectral remote sensing image data using UAV aerial surveys, establishes a database of rural NPG water bodies, and conducts software simulations and model building to achieve rapid identification and monitoring of rural water pollution. This technology can dynamically monitor changes in the water environment over a large spatial area and quickly track sudden environmental pollution events, thus providing a basis for environmental governance decisions. However, this type of remote sensing-based monitoring technology still has certain limitations. For example, remote sensing data is greatly affected by weather conditions, spatial resolution, and water body obstruction, making it difficult to conduct high-frequency continuous monitoring of small-scale rural water bodies. Furthermore, the quantitative accuracy of remote sensing methods for internal chemical indicators (such as total phosphorus content) is limited, making it difficult to achieve real-time and precise assessment of the degree of water pollution.

[0005] To address this issue, this invention proposes an edge IoT method and system for monitoring total phosphorus content in rural water bodies. By utilizing a lightweight deep inference model to assess total phosphorus content at the edge, and combining collaborative communication and time slot scheduling mechanisms between edge nodes, it achieves low-energy, high-efficiency transmission of multiple data packets, enabling continuous, stable, and accurate monitoring of total phosphorus content in rural water bodies, and providing reliable data support for rural water environment management and pollution early warning. Summary of the Invention

[0006] This invention provides a method and system for monitoring total phosphorus content in rural water bodies using edge IoT. Traditional monitoring methods rely on manual sampling or centralized computation, resulting in slow response times and difficulty in handling large-scale edge data. Step S1, by deploying multi-mode sensors at edge nodes, utilizing a heterogeneous signal fusion framework, and combining adaptive correction processing, effectively integrates environmental and water quality data, overcoming the bias of single sensors and achieving high-precision edge sensing data acquisition. Step S2 utilizes a lightweight deep inference model to extract features and assess total phosphorus content from the sensing data, enabling rapid, real-time assessment of total phosphorus content at edge nodes, ensuring low latency and edge computability in data processing, while reducing the computational burden on the cloud. It also achieves high-precision estimation even in the absence of direct chemical detection equipment. Step S3 performs joint optimization scheduling based on a collaborative protocol and time slot allocation algorithm, dynamically adjusting transmission strategies and time slot allocation to achieve a balance between bandwidth and energy consumption, improving the transmission stability of edge nodes. Step S4 verifies and adaptively corrects total phosphorus content by fusing historical transmission data, combining upstream and downstream node data, and using edge monitoring correction methods based on spatial consistency and temporal continuity, achieving automatic correction of abnormal data and improving the reliability and continuity of total phosphorus content monitoring.

[0007] To achieve the above objectives, this invention provides a method for monitoring total phosphorus content in rural water bodies using edge IoT, comprising the following steps: S1: Deploy multi-mode sensors at the edge nodes of rural water bodies, and synchronously collect environmental data and water quality data based on a heterogeneous signal fusion framework. Use a microprocessor at the edge node to adaptively correct the collected environmental data and water quality data to form sensing data of rural water bodies at the edge node locations. S2: Use a lightweight deep inference model to extract features and assess the total phosphorus content of the rural water body's perception data. The edge nodes will construct a transmission data packet based on the assessed total phosphorus content and the location of the edge nodes. S3: Based on the collaborative protocol and time slot allocation algorithm between the edge nodes, the bandwidth and energy consumption of the edge nodes are jointly optimized and scheduled. According to the joint optimization and scheduling results, the transmission strategy of data packets between the edge nodes is dynamically adjusted. According to the dynamically adjusted transmission strategy, the edge nodes transmit the data packets to the cloud platform. S4: Based on the transmission data packets of the edge nodes, merge the historical transmission data packets of the edge nodes, use the edge monitoring correction method to verify and adaptively correct the total phosphorus content in the transmission data packets, and output the corrected total phosphorus content.

[0008] As a further improvement of the present invention: Furthermore, in step S1, multi-mode sensors are deployed at the edge nodes of rural water bodies to synchronously collect environmental and water quality data based on a heterogeneous signal fusion framework, including: S11: Select the upstream inlet, midstream slow-flow zone, downstream outlet, and tributary confluence of rural waterways to deploy edge nodes, and deploy multi-mode sensors at the edge node locations. The multi-mode sensors are integrated from pH sensor, dissolved oxygen sensor, turbidity sensor, conductivity sensor, water temperature sensor, air temperature sensor, relative humidity sensor, light intensity sensor, and wind speed sensor. S12: The multimode sensor periodically and synchronously collects environmental data and water quality data based on a heterogeneous signal fusion framework. The heterogeneous signal fusion framework means that all sensors in the multimode sensor are set with a unified time reference and the same sampling frequency, and the data collected by different sensors are normalized in terms of dimensions.

[0009] Furthermore, step S1, which uses a microprocessor at the edge node to adaptively correct the collected environmental and water quality data, also includes: S13: Based on the historical data collected by each sensor in the multi-mode sensor, a confidence coefficient for each sensor in the multi-mode sensor is constructed. Based on the confidence coefficient, the collected data of each sensor in the environmental data and water quality data are simultaneously corrected for deviation to obtain the environmental data and water quality data after deviation correction. The historical data includes the mean of historical data collected by the sensor, the standard deviation of historical data collected, and the historical data collected error. S14: Generate the drift estimation sequence of the bias-corrected environmental data and water quality data, and perform drift self-calibration processing on the bias-corrected environmental data and water quality data to obtain drift self-calibrated environmental data and water quality data. S15: Generate dynamic thresholds for the environmental data and water quality data after drift self-calibration processing, and based on the dynamic thresholds, filter out data values ​​in the environmental data and water quality data after drift self-calibration processing that do not meet the dynamic threshold requirements, compress and correct the filtered data values ​​to obtain compressed and corrected environmental data and water quality data, and use the compressed and corrected environmental data and water quality data as the sensing data of rural water bodies at the edge node location.

[0010] Furthermore, in step S2, a lightweight deep inference model is used to extract features and assess the total phosphorus content of the perceived data of the rural water body, including: S21: The lightweight deep inference model includes an input layer, a temporal feature encoding layer, a gated update layer, and a lightweight regression layer. The input layer is used to receive the perception data of rural water bodies and convert the perception data of rural water bodies into a perception matrix. S22: The temporal feature encoding layer uses a depthwise separable convolutional structure to extract temporal features from the perception matrix, thereby obtaining the temporal features corresponding to the perception matrix; S23: The gated update layer updates the temporal state of the temporal features based on the gating mechanism, and generates the temporal state vector corresponding to the temporal features; S24: The lightweight regression layer uses a regression method to evaluate the total phosphorus content of the time-series state vector, and obtains the total phosphorus content corresponding to the perceived data of the rural water body.

[0011] Furthermore, in step S2, the edge nodes construct a transmission data packet using the assessed total phosphorus content and the edge node's location, and this step also includes: The edge node extracts the total phosphorus content and the timestamp of the total phosphorus content generation, and concatenates the total phosphorus content, the timestamp of the total phosphorus content generation, and the position of the edge node into a transmission data vector, generates a check value for the transmission data vector, and constructs a transmission data packet from the transmission data vector and the check value.

[0012] Furthermore, in step S3, based on the cooperation protocol and time slot allocation algorithm between the edge nodes, a joint optimization scheduling is performed on the transmission mode, transmission priority, and transmission time slot length of the edge nodes, including: S31: The edge node collects its own real-time communication status parameters and exchanges the real-time communication status parameters through the neighborhood broadcast mechanism to generate a network status matrix containing the real-time communication status parameters of all edge nodes. The real-time communication status parameters include available bandwidth, remaining power, and communication distance between the edge node and the cloud platform. S32: Calculate the node priority scheduling index of the edge node using the joint optimization scheduling function, and generate the transmission mode of the edge node based on the node priority scheduling index, wherein the transmission mode of the edge node includes direct transmission and cooperative transmission. Specifically, the calculation formula for the joint optimization scheduling function is as follows: ; in, This represents the joint optimization scheduling function. This represents the real-time communication status parameters of the m-th edge node, where M represents the number of edge nodes. This represents the node priority scheduling index for the m-th edge node. These represent the available bandwidth, remaining power, and communication distance to the cloud platform for the m-th edge node, respectively. These represent the normalized reference maximum values ​​for available bandwidth, remaining battery power, and communication distance between the edge node and the cloud platform, respectively. These represent the scheduling coefficients for available bandwidth, remaining power, and communication distance between the edge node and the cloud platform, respectively. The higher the node priority scheduling index, the better the communication conditions for data transmission at the edge node. Based on the node priority scheduling index, the transmission mode of the edge node is generated. An example of generating the transmission mode for the m-th edge node is as follows: If If the probability exceeds a preset probability threshold, the transmission mode of the m-th edge node is direct transmission; otherwise, the transmission mode of the m-th edge node is cooperative transmission. The collaborative transmission means that the edge node prioritizes transmitting data packets to the nearest neighboring edge node with a higher node priority scheduling index, and the neighboring edge node then transmits the data packets to the cloud platform. S33: Convert the node priority scheduling index into the transmission priority of the edge node, sort the edge nodes in ascending order of transmission priority, and generate the transmission time slot length of the edge node.

[0013] Furthermore, step S3, which dynamically adjusts the data packet transmission strategy between the edge nodes based on the joint optimization scheduling result, also includes: The edge nodes transmit the transmission data packets constructed by the edge nodes in the order based on transmission priority, transmission mode, and transmission time slot length. If a transmitted data packet needs to be retransmitted, the transmission process of the next data packet is automatically postponed, and the retransmission of the data packet is given priority.

[0014] Furthermore, in step S4, the historical transmission data packets of the edge nodes are fused, and the total phosphorus content in the transmission data packets is verified and adaptively corrected using an edge monitoring correction method, including: S41: The cloud platform parses the transmission data packets of the edge node, extracts the latitude and longitude coordinates, total phosphorus content, and timestamp of the total phosphorus content generation of the edge node, and obtains the most recently successfully transmitted historical transmission data packets of the edge node, extracting the adaptive correction result of the total phosphorus content in the historical transmission data packets as the historical total phosphorus content of the edge node location. S42: Based on the water flow direction of the rural water body and the latitude and longitude coordinates of the edge node, filter the nearest upstream node and the nearest downstream node of the edge node from the edge node set. The edge node set contains all edge nodes. The nearest upstream node is the edge node in the edge node set that is located upstream of the edge node and is closest to the edge node. The nearest downstream node is the edge node in the edge node set that is located downstream of the edge node and is closest to the edge node. S43: Construct a total phosphorus content verification function based on spatial consistency and temporal continuity to verify the total phosphorus content of the edge nodes, wherein the expression of the total phosphorus content verification function is: ; in, This represents the function for verifying total phosphorus content. This represents the verification information of the m-th edge node, where M represents the number of edge nodes. The verification information includes the total phosphorus content of the nearest upstream and nearest downstream nodes of the edge node, the historical total phosphorus content, and the total phosphorus content of the edge node itself. This represents the total phosphorus content verification function value of the m-th edge node. This represents the total phosphorus content of the m-th edge node. This represents the total phosphorus content of the nearest upstream neighbor of an edge node. This represents the total phosphorus content of the nearest downstream node of an edge node. represents the upstream node weight, represents the historical total phosphorus content of the m-th edge node, and represents the time continuity verification coefficient; like If the total phosphorus content of the m-th edge node exceeds the preset phosphorus content variation threshold, it indicates that there is an error in the total phosphorus content, and adaptive correction will be performed.

[0015] This invention also proposes an edge IoT system for monitoring total phosphorus content in rural water bodies. The edge IoT system for monitoring total phosphorus content in rural water bodies includes a multi-mode sensor, an edge node, and a cloud platform. The edge node includes a communication module, a power supply unit, a microprocessor, and a lightweight inference module. The lightweight inference module has a built-in lightweight deep inference model. The power supply unit is powered by a solar panel and a battery to realize the edge IoT method for monitoring total phosphorus content in rural water bodies as described above.

[0016] Compared with existing technologies, this invention proposes a method and system for monitoring total phosphorus content in rural water bodies using edge IoT, which has the following beneficial effects: First, this invention constructs a lightweight deep inference model comprising a temporal feature encoding layer, a gated update layer, and a lightweight regression layer to achieve efficient feature extraction and total phosphorus content assessment from multi-source sensing data of rural water bodies. Specifically, the depthwise separable convolutional structure reduces computational complexity and extracts key temporal features, the gated update mechanism adaptively filters effective information and suppresses noise interference, and the lightweight regression layer enables rapid estimation of total phosphorus content. This significantly reduces the computational load on edge nodes while maintaining prediction accuracy, thereby improving the real-time performance and energy efficiency of online monitoring of rural water bodies.

[0017] Meanwhile, this invention achieves multi-source consistency verification of the currently assessed total phosphorus content by constructing a total phosphorus content verification function that includes upstream nodes, downstream nodes, and historical monitoring data. This is used to characterize the difference between the total phosphorus content of the current edge node and the weighted average of upstream and downstream nodes. Since the total phosphorus content in water usually exhibits continuous diffusion or gradual change along the water flow direction, the evaluation results between adjacent edge nodes generally do not show abrupt changes. Spatial reference values ​​are constructed using the total phosphorus content of upstream and downstream nodes. This system can determine whether the current node monitoring results conform to the spatial distribution pattern of water flow, thereby effectively identifying abnormal values ​​caused by local sensor drift or instantaneous noise. By introducing historical total phosphorus content as a time reference benchmark, it can avoid data mutations caused by instantaneous communication errors or short-term sensor anomalies. This allows the total phosphorus content verification function to comprehensively evaluate the rationality of the data from the perspectives of water spatial propagation patterns and monitoring time continuity. It can effectively identify abnormal assessment results without adding additional sensing equipment, thereby improving the data reliability and monitoring stability of the entire edge IoT water monitoring system. Attached Figure Description

[0018] Figure 1 This is a schematic flowchart of a method for monitoring total phosphorus content in rural water bodies using edge IoT, provided in an embodiment of the present invention. Figure 2 This is a schematic diagram of a data packet transmission strategy for multiple sets of edge nodes provided in an embodiment of the present invention; Figure 3 This is a comparison chart of total phosphorus content provided in an embodiment of the present invention.

[0019] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0020] The realization of the objectives, functional characteristics, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0021] This invention provides a method for monitoring total phosphorus content in rural water bodies using edge IoT. The executing entity of this method includes, but is not limited to, at least one electronic device configured to execute the method provided in this invention, such as a server or a terminal. In other words, the method can be executed by software or hardware installed on a terminal device or a server device, where the software may be a blockchain platform. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster.

[0022] Reference Figure 1 as well as Figure 2 As shown, Embodiment 1 of the present invention is as follows: A method for monitoring total phosphorus content in rural water bodies using edge IoT, the method comprising: S1: Deploy multi-mode sensors at the edge nodes of rural water bodies, and synchronously collect environmental data and water quality data based on a heterogeneous signal fusion framework. Use a microprocessor at the edge node to adaptively correct the collected environmental data and water quality data to form sensing data of rural water bodies at the edge node locations.

[0023] Specifically, multi-mode sensors are deployed at the edge nodes of rural water bodies to simultaneously collect environmental and water quality data based on a heterogeneous signal fusion framework, including: S11: Select the upstream inlet, midstream slow-flow zone, downstream outlet, and tributary confluence of rural waterways to deploy edge nodes, and deploy multi-mode sensors at the edge node locations. The multi-mode sensors are integrated from pH sensor, dissolved oxygen sensor, turbidity sensor, conductivity sensor, water temperature sensor, air temperature sensor, relative humidity sensor, light intensity sensor, and wind speed sensor. S12: The multimode sensor periodically and synchronously collects environmental data and water quality data based on a heterogeneous signal fusion framework. The heterogeneous signal fusion framework means that all sensors in the multimode sensor are set with a unified time reference and the same sampling frequency, and the data collected by different sensors are normalized in terms of dimensions.

[0024] It should be noted that the multi-mode sensor periodically collects environmental data and water quality data, and the collected environmental data and water quality data are in the form of fixed-length and time-synchronized sequence data. Specifically, a connection between multi-mode sensors and edge nodes is established using sensing probes and a data acquisition module. The edge nodes are installed on corrosion-resistant supports or buoy platforms. The sensing probes include two types, which are connected to environmental sensors and water quality sensors, respectively, and are located above and inside the rural river channel, respectively. The environmental sensors include air temperature sensors, relative humidity sensors, light intensity sensors, and wind speed sensors. The water quality sensors include pH sensors, dissolved oxygen sensors, turbidity sensors, conductivity sensors, and water temperature sensors. All sensors are connected to the data acquisition module via an RS485 interface. The analog-to-digital conversion unit in the data acquisition module samples the data to form environmental data and water quality data, which are then transmitted to the edge nodes.

[0025] Step S1 uses microprocessors at edge nodes to adaptively correct the collected environmental and water quality data, and also includes: S13: Based on the historical data collected by each sensor in the multi-mode sensor, a confidence coefficient for each sensor in the multi-mode sensor is constructed. Based on the confidence coefficient, the collected data of each sensor in the environmental data and water quality data are simultaneously corrected for deviation to obtain the environmental data and water quality data after deviation correction. The historical data includes the mean of historical data collected by the sensor, the standard deviation of historical data collected, and the historical data collected error. Specifically, the historical acquisition data refers to the sensor's acquisition data in historical periods. All historical acquisition data are normalized in terms of dimensions. During the sensor's data acquisition process, acquisition data is randomly selected for error measurement, thereby updating the sensor's historical acquisition error in real time. The historical acquisition error is the average error between the previously randomly selected Len acquisition data and the actual measurement data. The default setting for Len is 50. As an embodiment of the present invention, the deviation correction formula for the sensor's acquired data is as follows: ; ; in, Indicates the first Data collected by each sensor Indicates collected data The deviation correction results Indicates the first Confidence coefficient of each sensor, Indicates the first Historical acquisition errors of each sensor Indicates the first The historical data standard deviation of each sensor Indicates the first The average historical data collected by each sensor, where K represents the number of sensors. The data collected by the sensors are the data collected by the sensors at the sampling time. The data collected at continuous sampling times constitute environmental data and water quality data. S14: Generate the drift estimation sequence of the bias-corrected environmental data and water quality data, and perform drift self-calibration processing on the bias-corrected environmental data and water quality data to obtain drift self-calibrated environmental data and water quality data. Specifically, the environmental data and water quality data after deviation correction are represented in the following forms: ; Where y represents the environmental data and water quality data after bias correction, This represents the j-th sequence data in the environmental data and water quality data y after bias correction. The 1st to 9th sequence data are, respectively, environmental temperature data, environmental relative humidity data, environmental light intensity data, environmental wind speed data, water quality pH data, water quality dissolved oxygen data, water quality turbidity data, water quality conductivity data, and water quality temperature data. Represents sequence data The first to Nth data values, where N represents the data length of the collected environmental and water quality data; The ambient temperature data, ambient relative humidity data, ambient light intensity data, and ambient wind speed data are environmental data, which are collected sequentially by a temperature sensor, a relative humidity sensor, a light intensity sensor, and a wind speed sensor, respectively. The water quality data, including pH, dissolved oxygen, turbidity, conductivity, and temperature, are water quality data, which are collected sequentially using a pH sensor, a dissolved oxygen sensor, a turbidity sensor, a conductivity sensor, and a temperature sensor, respectively. The sequence data The drift estimation sequence generation result is as follows: ; ; in, Represents the sequence data The drift estimation sequence, This represents the N sequence values ​​of the drift estimation sequence. Represents the drift estimation sequence The nth sequence value in the sequence, , Represents the sequence update coefficients, set It is 0.3. Represents sequence data The nth data value in Represents sequence data The mean of the data values; specifically, the default setting. ; Based on the drift estimation sequence pair sequence data The data values ​​in the data are subjected to drift self-calibration processing, where the formula for drift self-calibration processing is: , where represents the data value The result of the drift self-calibration process is then... Represents sequence data The drift self-calibration processing results express The first to Nth data values ​​in the dataset; S15: Generate dynamic thresholds for the environmental data and water quality data after drift self-calibration processing, and based on the dynamic thresholds, filter out data values ​​in the environmental data and water quality data after drift self-calibration processing that do not meet the dynamic threshold requirements, compress and correct the filtered data values ​​to obtain compressed and corrected environmental data and water quality data, and use the compressed and corrected environmental data and water quality data as the sensing data of rural water bodies at the edge node location.

[0026] Specifically, for any sequence data in the environmental data and water quality data after the drift self-calibration process. Generate this sequence data dynamic threshold ,in This represents the threshold coefficient; the default setting is 3. Represents sequence data Standard deviation; According to the dynamic threshold The sequence data Middle data value Examples that meet the dynamic threshold requirements are: , where represents sequence data The mean of the data values If the dynamic threshold requirement is not met, that is, for the data value... Compression correction is performed, and the formula for compression correction is: ; The representation of the sensing data of the rural water bodies is as follows: ; in, This represents the sensing data of rural water bodies. Sensing data representing rural water bodies The j-th sequence data, Represents sequence data The first to Nth data values.

[0027] It should be noted that this invention utilizes historical acquisition mean, historical acquisition standard deviation, and historical acquisition error to construct sensor confidence coefficients, and applies weighted bias correction to the acquired data of each sensor. This gives greater weight to data from sensors with higher reliability, while data from sensors with poor stability or large errors automatically align with historical statistical characteristics, thereby reducing the impact of individual sensor measurement errors on the overall sensing data. By constructing a drift estimation sequence and performing drift self-calibration on the data, zero-point drift or slow systematic shifts generated by sensors during long-term acquisition and operation can be dynamically estimated, ensuring that the data remains within a stable statistical distribution range. This achieves adaptive compensation for systematic drift and improves the reliability of long-term monitoring. By constructing a dynamic threshold based on standard deviation, data that does not meet the statistical stability constraints based on the dynamic threshold are compressed and corrected, effectively suppressing extreme data caused by instantaneous outliers or environmental mutations. This avoids outliers from disrupting the overall data distribution and amplifying their impact on subsequent model inference and water quality assessment results, thus improving data stability. Through the aforementioned multi-layered data correction mechanism, adaptive correction of environmental and water quality data can be achieved at the edge node side, improving the stability, robustness, and reliability of rural water monitoring data, while reducing the probability of abnormal data being propagated to the cloud platform.

[0028] S2: Use a lightweight deep inference model to extract features and assess the total phosphorus content of the perceived data of the rural water body. The edge node constructs the assessed total phosphorus content and the position of the edge node into a transmission data packet.

[0029] Specifically, a lightweight deep inference model is used to extract features and assess the total phosphorus content of the perceived data of the rural water bodies, including: S21: The lightweight deep inference model includes an input layer, a temporal feature encoding layer, a gated update layer, and a lightweight regression layer. The input layer is used to receive the perception data of rural water bodies and convert the perception data of rural water bodies into a perception matrix. Specifically, the sensing data of the rural water bodies The corresponding perception matrix form is: ; Wherein, the sensing data of the rural water bodies is represented. The corresponding perception matrix, where T represents the transpose. The data represent the sensing data of rural water bodies in the following order. Sequence data from type 1 to type 9; S22: The temporal feature encoding layer uses a depthwise separable convolutional structure to extract temporal features from the perception matrix, thereby obtaining the temporal features corresponding to the perception matrix; S23: The gated update layer updates the temporal state of the temporal features based on the gating mechanism, and generates the temporal state vector corresponding to the temporal features; Specifically, the formula for generating the time-series state vector is: ; ; ; in, Represents the perception matrix The corresponding time series features, Represents the temporal state vector. Indicates the gating coefficient. This indicates the activation function; the default activation function is the Sigmoid function. This represents the trainable convolutional matrix parameters in the gated update layer. This represents the trainable bias parameters in the gated update layer. Representing time series features The hidden state update result represents element-wise multiplication. Indicate length and Consistent vectors of all 1s The lengths are the same. This represents the hyperbolic tangent function; specifically, the temporal features and temporal state vectors are both in 9-row, 1-column vector form. S24: The lightweight regression layer uses a regression method to evaluate the total phosphorus content of the time-series state vector, and obtains the total phosphorus content corresponding to the perceived data of the rural water body.

[0030] Specifically, the timing state vector The regression calculation formula is: ; in, This represents the vector of trainable regression coefficients in a lightweight regression layer. This represents the trainable bias parameters in a lightweight regression layer. The sensing data representing the rural water bodies The corresponding total phosphorus content, where the total phosphorus content is in scalar form.

[0031] As an embodiment of the present invention, by simultaneously collecting the actual measured values ​​of total phosphorus content and sensing data of different rural water bodies, the total phosphorus content corresponding to the sensing data is output using the lightweight deep inference model. With the goal of minimizing the absolute value of the difference between the actual measured value of total phosphorus content and the total phosphorus content output by the model, a training loss function for the trainable parameters in the lightweight deep inference model is constructed. Based on the training loss function, the gradient descent algorithm or the Adam optimizer is used to optimize and train the trainable parameters in the lightweight deep inference model.

[0032] The edge nodes construct a transmission data packet based on the assessed total phosphorus content and the location of the edge nodes, and also include: The edge node extracts the total phosphorus content and the timestamp of the total phosphorus content generation, and concatenates the total phosphorus content, the timestamp of the total phosphorus content generation, and the position of the edge node into a transmission data vector, generates a check value for the transmission data vector, and constructs a transmission data packet from the transmission data vector and the check value.

[0033] Specifically, the formula for calculating the verification value is as follows: ; Where C represents the check value, represents the transmitted data vector, represents the sum of the transmitted data vectors, represents the modulo operator, and represents the check modulus, which is set to 256 by default. As an embodiment of the present invention, when the cloud platform receives the transmission data packet transmitted by the edge node, the cloud platform extracts the transmission data vector in the transmission data packet and recalculates the check value. When the recalculated check value is consistent with the check value in the transmission data packet, it indicates that the transmission data packet has passed the check and the data in the transmission data packet has not been tampered with. Otherwise, the transmission data packet needs to be retransmitted.

[0034] S3: Based on the collaborative protocol and time slot allocation algorithm between the edge nodes, the bandwidth and energy consumption of the edge nodes are jointly optimized and scheduled. According to the joint optimization and scheduling results, the transmission strategy of data packets between the edge nodes is dynamically adjusted. According to the dynamically adjusted transmission strategy, the edge nodes transmit data packets to the cloud platform.

[0035] Specifically, based on the cooperation protocol and time slot allocation algorithm between the edge nodes, the transmission mode, transmission priority, and transmission time slot length of the edge nodes are jointly optimized and scheduled, including: S31: The edge node collects its own real-time communication status parameters and exchanges the real-time communication status parameters through the neighborhood broadcast mechanism to generate a network status matrix containing the real-time communication status parameters of all edge nodes. The real-time communication status parameters include available bandwidth, remaining power, and communication distance between the edge node and the cloud platform. S32: Calculate the node priority scheduling index of the edge node using the joint optimization scheduling function, and generate the transmission mode of the edge node based on the node priority scheduling index, wherein the transmission mode of the edge node includes direct transmission and cooperative transmission. Specifically, the calculation formula for the joint optimization scheduling function is as follows: ; in, This represents the joint optimization scheduling function. This represents the real-time communication status parameters of the m-th edge node. This represents the node priority scheduling index for the m-th edge node. , These represent the available bandwidth, remaining power, and communication distance to the cloud platform for the m-th edge node, respectively. These represent the normalized reference maximum values ​​for available bandwidth, remaining battery power, and communication distance between the edge node and the cloud platform, respectively. The scheduling coefficients, representing available bandwidth, remaining power, and communication distance between the edge node and the cloud platform, are set in sequence. The values ​​are 0.5, 0.25, and 0.25, respectively. The default settings for the normalized reference maximum values ​​of available bandwidth, remaining power, and communication distance between edge nodes and the cloud platform are 10Mbps, 5000mAh, and 5000 meters, respectively. The higher the node priority scheduling index, the better the communication conditions for data transmission at the edge node. Based on the node priority scheduling index, the transmission mode of the edge node is generated. An example of generating the transmission mode for the m-th edge node is as follows: If If the probability exceeds a preset threshold (default setting is 0.3), the transmission method of the m-th edge node is direct transmission; otherwise, the transmission method of the m-th edge node is cooperative transmission. The collaborative transmission means that the edge node prioritizes transmitting data packets to the nearest neighboring edge node with a higher node priority scheduling index, and the neighboring edge node then transmits the data packets to the cloud platform. S33: Convert the node priority scheduling index into the transmission priority of the edge node, sort the edge nodes in ascending order of transmission priority, and generate the transmission time slot length of the edge node.

[0036] Specifically, the conversion formula for the transmission priority is: ; in, This represents the transmission priority of the m-th edge node. A higher node priority scheduling index corresponds to a lower transmission priority, indicating that the data packets transmitted within the edge node arrive earlier in the transmission sequence. Represents a set The minimum value in the index represents the minimum node priority scheduling index among all edge nodes. Represents a set The maximum value in the value represents the highest node priority scheduling index among all edge nodes; As an embodiment of the present invention, the formula for calculating the transmission time slot length of the edge node is: ; in, Indicates the transmission time slot length of the edge node. Indicates the duration of the communication cycle; default setting. It lasts for 60 seconds. This represents the transmission adjustment coefficient, which is set to 0.5 by default. The higher the node priority scheduling index, the longer the transmission time slot length of the edge node, ensuring that the edge node can transmit to the cloud platform in the direct transmission mode and reserving retransmission time.

[0037] It should be noted that this invention achieves coordinated scheduling management of bandwidth resources and energy consumption status of edge nodes by constructing a joint optimization scheduling mechanism based on real-time communication status parameters. Specifically, this invention uses a joint optimization scheduling function to perform normalized and fused calculations on available bandwidth, remaining power, and communication distance to generate node priority scheduling indicators. This ensures communication efficiency while taking into account node energy consumption, achieving a dynamic balance between network communication resources and node lifetime. Furthermore, based on the node priority scheduling indicators, it adaptively generates direct transmission or cooperative transmission modes, allowing edge nodes with better communication conditions to prioritize data packet transmission tasks, while nodes with weaker communication conditions forward data packets through neighboring nodes. This effectively reduces energy consumption and packet loss risks associated with long-distance transmission.

[0038] Furthermore, by mapping node priority scheduling metrics to transmission priorities and sorting them in ascending order, and combining this with the communication cycle to generate differentiated transmission time slot lengths, nodes with better communication conditions are given more transmission time and reserving retransmission space. This reduces data collisions and channel contention, thereby improving network transmission stability and throughput efficiency. Simultaneously, this scheduling mechanism can complete transmission strategies and time slot allocation based solely on the calculated node priority scheduling metrics, reducing the computational burden on edge nodes and improving the overall system's real-time response capability and energy efficiency.

[0039] Step S3, which dynamically adjusts the data packet transmission strategy between edge nodes based on the joint optimization scheduling result, further includes: The edge nodes transmit the transmission data packets constructed by the edge nodes in the order based on transmission priority, transmission mode, and transmission time slot length. If a transmitted data packet needs to be retransmitted, the transmission process of the next data packet is automatically postponed, and the retransmission of the data packet is given priority.

[0040] Specifically, during transmission, if a data packet fails to reach the receiving end due to communication interference or network anomalies, the transmission process of the next data packet will be automatically paused, and the failed data packet will be retransmitted first. During retransmission, the edge node maintains the original transmission time slot length and transmission mode, ensuring the effective utilization of node communication resources. By dynamically adjusting the transmission order and time slots, the edge node can simultaneously consider transmission reliability and energy consumption optimization, achieving low-power, high-reliability collaborative data upload.

[0041] For reference Figure 2 The diagram shows the data packet transmission strategy for multiple edge nodes. The earlier a data packet is in the sequence, the longer its transmission time slot. Data packets 1 and 2 in the diagram are transmitted directly, while the remaining data packets are transmitted collaboratively.

[0042] S4: Based on the transmission data packets of the edge nodes, merge the historical transmission data packets of the edge nodes, use the edge monitoring correction method to verify and adaptively correct the total phosphorus content in the transmission data packets, and output the corrected total phosphorus content.

[0043] In step S4, historical transmission data packets from edge nodes are fused, and the total phosphorus content in the transmission data packets is verified and adaptively corrected using an edge monitoring correction method, including: S41: The cloud platform parses the transmission data packets of the edge node, extracts the latitude and longitude coordinates, total phosphorus content, and timestamp of the total phosphorus content generation of the edge node, and obtains the most recently successfully transmitted historical transmission data packets of the edge node, extracting the adaptive correction result of the total phosphorus content in the historical transmission data packets as the historical total phosphorus content of the edge node location. S42: Based on the water flow direction of the rural water body and the latitude and longitude coordinates of the edge node, filter the nearest upstream node and the nearest downstream node of the edge node from the edge node set. The edge node set contains all edge nodes. The nearest upstream node is the edge node in the edge node set that is located upstream of the edge node and is closest to the edge node. The nearest downstream node is the edge node in the edge node set that is located downstream of the edge node and is closest to the edge node. S43: Construct a total phosphorus content verification function based on spatial consistency and temporal continuity to verify the total phosphorus content of the edge nodes, wherein the expression of the total phosphorus content verification function is: ; in, This represents the function for verifying total phosphorus content. This represents the verification information of the m-th edge node, where M represents the number of edge nodes. The verification information includes the total phosphorus content of the nearest upstream and nearest downstream nodes of the edge node, the historical total phosphorus content, and the total phosphorus content of the edge node itself. This represents the total phosphorus content verification function value of the m-th edge node. Let represent the total phosphorus content of the m-th edge node, and let represent the total phosphorus content of the nearest upstream neighbor of the edge node. This represents the total phosphorus content of the nearest downstream node of an edge node. This indicates the weight of the upstream node (default setting is 0.6). This represents the historical total phosphorus content of the m-th edge node. This represents the time continuity check coefficient, the default setting. It is 0.2; like If the total phosphorus content exceeds the preset phosphorus content variation threshold, it indicates that the total phosphorus content of the m-th edge node may be affected by sensor drift or communication noise, and adaptive correction is required.

[0044] Specifically, the preset threshold for phosphorus content variation is set to 0.1 mg / L; As an embodiment of the present invention, if the total phosphorus content of the m-th edge node needs to be adaptively corrected, the adaptive correction formula is as follows: ; in, Indicates total phosphorus content The adaptive correction results Indicates spatial correction weights, default settings. With values ​​of 0 and 2, this adaptive correction formula can suppress abnormal deviations while preserving the original total phosphorus content assessment results, making the correction results more consistent with the concentration change pattern along the water body's flow direction, thereby improving the stability and reliability of total phosphorus content monitoring data.

[0045] Example 2: An edge IoT monitoring system for total phosphorus content in rural water bodies includes a multi-mode sensor, an edge node, and a cloud platform. The edge node includes a communication module, a power supply unit, a microprocessor, and a lightweight inference module. The lightweight inference module has a built-in lightweight deep inference model. The power supply unit is powered by a solar panel and a battery, forming a long-term unattended operation mode. Multimode sensors are used to synchronously acquire environmental and water quality data based on a heterogeneous signal fusion framework; Edge nodes are used to adaptively correct the collected environmental and water quality data to form perception data of rural water bodies at the edge node locations. A lightweight deep inference model is used to extract features and assess the total phosphorus content of the perception data of rural water bodies. The assessed total phosphorus content and the location of the edge nodes are used to construct a transmission data packet. The microprocessor performs joint optimization scheduling of the bandwidth and energy consumption of the edge nodes. Based on the joint optimization scheduling results, the transmission strategy of the transmission data packets between the edge nodes is dynamically adjusted. According to the dynamically adjusted transmission strategy, the edge nodes transmit the transmission data packets to the cloud platform. The cloud platform is used to integrate historical data transmission data of edge nodes, and uses edge monitoring correction method to verify and adaptively correct the total phosphorus content in the data transmission data, and outputs the corrected total phosphorus content to realize the edge IoT rural water body total phosphorus content monitoring method as described in Example 1.

[0046] Example 3: As another embodiment of the present invention, for the edge monitoring correction method described in step S4, four groups of rural water body locations with different positions are selected, and the total phosphorus content evaluated at each selected rural water body location needs to undergo adaptive correction processing as described in step S4. The true total phosphorus content is measured simultaneously, and the total phosphorus content at each rural water body location is evaluated as described in steps S1-S2. The true total phosphorus content, the evaluated total phosphorus content, and the adaptively corrected total phosphorus content value for each rural water body location are compared. The evaluated total phosphorus content value is the total phosphorus content evaluated in step S2, and the adaptively corrected total phosphorus content value is the adaptively corrected total phosphorus content output in step S4. (Refer to...) Figure 3 The total phosphorus content comparison chart shown shows that the adaptively corrected total phosphorus content is closer to the measured true total phosphorus content.

[0047] It should be understood that the embodiments described are for illustrative purposes only and are not limited to this structure in terms of the scope of the patent invention.

[0048] It should be noted that the sequence numbers of the above embodiments of the present invention are merely for descriptive purposes and do not represent the superiority or inferiority of the embodiments. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, apparatus, article, or method. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0049] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0050] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A method for monitoring total phosphorus content in rural water bodies using edge IoT, characterized in that, The method includes: S1: Deploy multi-mode sensors at the edge nodes of rural water bodies, and synchronously collect environmental data and water quality data based on a heterogeneous signal fusion framework. Use a microprocessor at the edge node to adaptively correct the collected environmental data and water quality data to form sensing data of rural water bodies at the edge node locations. S2: Use a lightweight deep inference model to extract features and assess the total phosphorus content of the rural water body's perception data. The edge nodes will construct a transmission data packet based on the assessed total phosphorus content and the location of the edge nodes. S3: Based on the collaborative protocol and time slot allocation algorithm between the edge nodes, the bandwidth and energy consumption of the edge nodes are jointly optimized and scheduled. According to the joint optimization and scheduling results, the transmission strategy of data packets between the edge nodes is dynamically adjusted. According to the dynamically adjusted transmission strategy, the edge nodes transmit the data packets to the cloud platform. S4: Based on the transmission data packets of the edge nodes, merge the historical transmission data packets of the edge nodes, use the edge monitoring correction method to verify and adaptively correct the total phosphorus content in the transmission data packets, and output the corrected total phosphorus content.

2. The method for monitoring total phosphorus content in rural water bodies using edge IoT as described in claim 1, characterized in that, In step S1, multi-mode sensors are deployed at the edge nodes of rural water bodies to synchronously collect environmental and water quality data based on a heterogeneous signal fusion framework, including: S11: Select the upstream inlet, midstream slow-flow zone, downstream outlet, and tributary confluence of rural waterways to deploy edge nodes, and deploy multi-mode sensors at the edge node locations. The multi-mode sensors are integrated from pH sensor, dissolved oxygen sensor, turbidity sensor, conductivity sensor, water temperature sensor, air temperature sensor, relative humidity sensor, light intensity sensor, and wind speed sensor. S12: The multimode sensor periodically and synchronously collects environmental data and water quality data based on a heterogeneous signal fusion framework. The heterogeneous signal fusion framework means that all sensors in the multimode sensor are set with a unified time reference and the same sampling frequency, and the data collected by different sensors are normalized in terms of dimensions.

3. The method for monitoring total phosphorus content in rural water bodies using edge IoT as described in claim 2, characterized in that, Step S1 uses microprocessors at edge nodes to adaptively correct the collected environmental and water quality data, and also includes: S13: Based on the historical data collected by each sensor in the multi-mode sensor, a confidence coefficient for each sensor in the multi-mode sensor is constructed. Based on the confidence coefficient, the collected data of each sensor in the environmental data and water quality data are simultaneously corrected for deviation to obtain the environmental data and water quality data after deviation correction. The historical data includes the mean of historical data collected by the sensor, the standard deviation of historical data collected, and the historical data collected error. S14: Generate the drift estimation sequence of the bias-corrected environmental data and water quality data, and perform drift self-calibration processing on the bias-corrected environmental data and water quality data to obtain drift self-calibrated environmental data and water quality data. S15: Generate dynamic thresholds for the environmental data and water quality data after drift self-calibration processing, and based on the dynamic thresholds, filter out data values ​​in the environmental data and water quality data after drift self-calibration processing that do not meet the dynamic threshold requirements, compress and correct the filtered data values ​​to obtain compressed and corrected environmental data and water quality data, and use the compressed and corrected environmental data and water quality data as the sensing data of rural water bodies at the edge node location.

4. The method for monitoring total phosphorus content in rural water bodies using edge IoT as described in claim 3, characterized in that, In step S2, a lightweight deep inference model is used to extract features and assess the total phosphorus content of the perceived data of the rural water body, including: S21: The lightweight deep inference model includes an input layer, a temporal feature encoding layer, a gated update layer, and a lightweight regression layer. The input layer is used to receive the perception data of rural water bodies and convert the perception data of rural water bodies into a perception matrix. S22: The temporal feature encoding layer uses a depthwise separable convolutional structure to extract temporal features from the perception matrix, thereby obtaining the temporal features corresponding to the perception matrix; S23: The gated update layer updates the temporal state of the temporal features based on the gating mechanism, and generates the temporal state vector corresponding to the temporal features; S24: The lightweight regression layer uses a regression method to evaluate the total phosphorus content of the time-series state vector, and obtains the total phosphorus content corresponding to the perceived data of the rural water body.

5. The method for monitoring total phosphorus content in rural water bodies using edge IoT as described in claim 4, characterized in that, In step S2, the edge nodes construct a transmission data packet based on the assessed total phosphorus content and the location of the edge nodes, and the process also includes: The edge node extracts the total phosphorus content and the timestamp of the total phosphorus content generation, and concatenates the total phosphorus content, the timestamp of the total phosphorus content generation, and the position of the edge node into a transmission data vector, generates a check value for the transmission data vector, and constructs a transmission data packet from the transmission data vector and the check value.

6. The method for monitoring total phosphorus content in rural water bodies using edge IoT as described in claim 1, characterized in that, In step S3, based on the cooperation protocol and time slot allocation algorithm between the edge nodes, a joint optimization scheduling is performed on the transmission mode, transmission priority, and transmission time slot length of the edge nodes, including: S31: The edge node collects its own real-time communication status parameters and exchanges the real-time communication status parameters through the neighborhood broadcast mechanism to generate a network status matrix containing the real-time communication status parameters of all edge nodes. The real-time communication status parameters include available bandwidth, remaining power, and communication distance between the edge node and the cloud platform. S32: Calculate the node priority scheduling index of the edge node using the joint optimization scheduling function, and generate the transmission mode of the edge node based on the node priority scheduling index, wherein the transmission mode of the edge node includes direct transmission and cooperative transmission. S33: Convert the node priority scheduling index into the transmission priority of the edge node, sort the edge nodes in ascending order of transmission priority, and generate the transmission time slot length of the edge node.

7. The method for monitoring total phosphorus content in rural water bodies using edge IoT as described in claim 1, characterized in that, Step S3, which dynamically adjusts the data packet transmission strategy between edge nodes based on the joint optimization scheduling result, further includes: The edge nodes transmit the transmission data packets constructed by the edge nodes in the order based on transmission priority, transmission mode, and transmission time slot length. If a transmitted data packet needs to be retransmitted, the transmission process of the next data packet is automatically postponed, and the retransmission of the data packet is given priority.

8. A method for monitoring total phosphorus content in rural water bodies using edge IoT as described in claim 7, characterized in that, In step S4, historical transmission data packets from edge nodes are fused, and the total phosphorus content in the transmission data packets is verified and adaptively corrected using an edge monitoring correction method, including: S41: The cloud platform parses the transmission data packets of the edge node, extracts the latitude and longitude coordinates, total phosphorus content, and timestamp of the total phosphorus content generation of the edge node, and obtains the most recently successfully transmitted historical transmission data packets of the edge node, extracting the adaptive correction result of the total phosphorus content in the historical transmission data packets as the historical total phosphorus content of the edge node location. S42: Based on the water flow direction of the rural water body and the latitude and longitude coordinates of the edge node, filter the nearest upstream node and the nearest downstream node of the edge node from the edge node set. The edge node set contains all edge nodes. The nearest upstream node is the edge node in the edge node set that is located upstream of the edge node and is closest to the edge node. The nearest downstream node is the edge node in the edge node set that is located downstream of the edge node and is closest to the edge node. S43: Construct a total phosphorus content verification function based on spatial consistency and temporal continuity to verify the total phosphorus content of the edge nodes, wherein the expression of the total phosphorus content verification function is: ； in, This represents the function for verifying total phosphorus content. This represents the verification information of the m-th edge node, where M represents the number of edge nodes. The verification information includes the total phosphorus content of the nearest upstream and nearest downstream nodes of the edge node, the historical total phosphorus content, and the total phosphorus content of the edge node itself. This represents the total phosphorus content verification function value of the m-th edge node. This represents the total phosphorus content of the m-th edge node. This represents the total phosphorus content of the nearest upstream neighbor of an edge node. This represents the total phosphorus content of the nearest downstream node of an edge node. Indicates the weight of the upstream node. This represents the historical total phosphorus content of the m-th edge node. Indicates the time continuity check coefficient; like If the total phosphorus content of the m-th edge node exceeds the preset phosphorus content variation threshold, it indicates that there is an error in the total phosphorus content, and adaptive correction will be performed.

9. An edge IoT system for monitoring total phosphorus content in rural water bodies, characterized in that, The edge IoT rural water body total phosphorus content monitoring system includes a multi-mode sensor, an edge node, and a cloud platform. The edge node includes a communication module, a power supply unit, a microprocessor, and a lightweight inference module. The lightweight inference module has a built-in lightweight deep inference model. The power supply unit is powered by a solar panel and a battery to realize the edge IoT rural water body total phosphorus content monitoring method as described in any one of claims 1-8.

Images