Flood season rainwater inlet sewage illegal discharge monitoring device based on internet of things and early warning method thereof

Abstract

The application discloses a flood season rainwater inlet sewage illegal discharge monitoring equipment based on Internet of Things and an early warning method thereof, relates to the technical field of intelligent hydrology and water quality sensing, and the early warning method is characterized in that monitoring equipment is installed at a river rainwater discharge port, historical flood season rainwater inlet rainfall, water level and water quality data are collected by using the monitoring equipment, a machine model is established by using the data, real-time monitoring data are collected and input into the trained model, water body conductivity is predicted, and according to a double-standard judgment method of the conductivity value, if the water is industrial wastewater, a water quality tracing model is immediately started to track the pollution source and report to a cloud platform. The application realizes real-time monitoring and early warning functions of flood season rainwater inlet industrial wastewater illegal discharge by combining Internet of Things technology and a water quality monitoring system, the water quality monitoring system can efficiently identify water quality changes and accurately locate the pollution source, and the accuracy of sewage tracing is greatly improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent hydrological and water quality sensing technology, specifically to an IoT-based monitoring device for illegal sewage discharge from storm drains during the flood season and its early warning method. Background Technology

[0002] Storm drains are an important component of drainage systems. Their primary function is to rapidly drain surface rainwater into the sewer system, preventing urban flooding, protecting cities from flooding disasters, and reducing damage to roads and buildings. Furthermore, storm drains are crucial for monitoring wastewater discharge during the flood season, preventing untreated pollutants from directly entering rivers and water bodies.

[0003] Currently, most storm drains rely on a combination of manual monitoring and online water quality monitoring equipment to obtain hydrological and water quality data. Manual monitoring is ineffective for real-time monitoring of illegal discharges and is costly in terms of manpower. Online water quality monitoring equipment is only used to detect basic water quality parameters and is typically installed only in key storm drains or high-risk pollution areas, making it impossible to trace the location of pollution. Summary of the Invention

[0004] This invention provides an IoT-based monitoring device and early warning method for illegal sewage discharge from storm drains during the flood season. It solves the problems mentioned in the background art, such as the poor real-time performance of obtaining hydrological and water quality information in traditional manual monitoring methods and the inability of online water quality monitoring equipment to trace pollution sources. This invention achieves the goal of real-time monitoring of hydrological and water quality information of storm drains and tracing the source of sewage discharge from storm drains.

[0005] This invention provides the following technical solution: an early warning method for monitoring sewage discharge from stormwater inlets during the flood season based on the Internet of Things (IoT). This method first installs monitoring equipment at river stormwater discharge outlets, collecting historical rainfall, water level, and water quality data from the stormwater inlets during the flood season. This data is then used to build a machine learning model. Real-time collected monitoring data is input into the trained model to predict water conductivity. Based on a dual-standard judgment method using conductivity values, if the water is industrial wastewater, a water quality source tracing model is immediately activated to track the pollution source and report it to a cloud platform. The monitoring equipment includes a waterproof casing, a hardware circuit board, a solar power supply system, hydrological sensors, disaster prevention and mitigation linkage sensors, agricultural drinking water sensors, a private cloud platform, a hydrological platform, and host computer software. The hardware circuit board includes a power module, a network module, a display and control module, a hydrological gateway module, a sensor configuration module, and sensor interfaces.

[0006] The early warning method includes the following steps:

[0007] S1. Model Establishment

[0008] Historical rainfall, water level, and water quality data from the flood season are used to train a machine learning model. The model input includes rainfall, water level change rate, and pH value, and the output is conductivity. This historical data is used to build a machine learning model.

[0009] S2, Data Acquisition

[0010] The monitoring equipment is installed around the rainwater outfall to collect rainfall, water level, and water quality data in real time and report them to the cloud platform. The hydrological sensors inside the monitoring equipment record the cumulative rainfall CP(t) and the periodic rainfall P(t), and the cumulative rainfall is periodically reset to prevent data overflow. They also record the water level Z(t) and are periodically calibrated to ensure measurement accuracy. The agricultural drinking water sensors measure pH value and conductivity.

[0011] S3, Dynamic Data Processing

[0012] During the operation of the water quality monitoring equipment, the system calculates the rainfall rate based on the real-time collected monitoring data. and rate of change of water level At the same time, the monitoring equipment will monitor the conductivity k of the water in real time through agricultural drinking water sensors. obs The standard deviation σ of the electrical conductivity of rainwater was calculated based on the time series.

[0013] S4, Model Input

[0014] The cumulative rainfall, rainfall over a period of time, rainfall rate, water level and its rate of change, and pH value data are input into a pre-trained machine learning model in real time. Based on the patterns learned during training, the model outputs the predicted conductivity of the water body. And predict the source strength Q;

[0015] S5. Anomaly Detection

[0016] The system employs risk-based tiered monitoring and adaptive sampling modes. Risk-based monitoring means that when the water conductivity is normal, the water quality monitoring system maintains a basic sampling frequency of once per hour; when the conductivity becomes abnormal, it automatically increases to a sampling frequency of once every 15 minutes. Adaptive sampling adjustment means that if the monitoring data changes significantly, even if no warning value is triggered, it will switch to high-frequency sampling, i.e., once per minute, and upload the data to the cloud platform. When the sensor detects high-intensity rainfall, the water quality monitoring system automatically activates the high-frequency sampling mode, and resumes low-frequency sampling after the data stabilizes, thereby optimizing energy efficiency.

[0017] A dual-indicator judgment standard is adopted: if Consider it normal rainwater; if and Consider it a minor abnormality, record and continue to observe, maintaining the baseline sampling frequency of one hour; if and Entering monitoring mode, the sampling frequency is increased to every 15 minutes; if and Treating this as a special case, and combining actual analysis and subsequent monitoring, assess whether there is a risk of illegal sewage discharge; if and The high-frequency sampling mode is activated, which means sampling once per minute and reporting to the water quality monitoring system to mark the possible illegal discharge of sewage, thus entering the source tracing analysis stage;

[0018] S6, Source Tracing Analysis

[0019] When the water quality monitoring system detects that the conductivity exceeds the standard, it immediately activates a multi-level source tracing algorithm. Through deep learning combined with traditional diffusion models, it gradually narrows down the range of pollution sources and accurately locates them.

[0020] The specific algorithm is as follows:

[0021] Step 1: Initially locate the pollution source range using an approximate optimization algorithm, and adjust the source strength Q using an optimization algorithm. j To reduce the sum of squared errors E, thereby identifying potential sewage discharge areas;

[0022] Step 2: In the initially identified potential pollution source areas, use a deep learning model to conduct a more detailed concentration distribution analysis. The input to the deep learning model includes Q obtained from the preliminary source tracing analysis. j Real-time monitoring data: rainfall, water level, conductivity, pH value. Based on the relative coordinates between each monitoring point and the pollution source, as well as the propagation path of pollutants in the water body, and the time series data of pollutant concentration changes over time, including the conductivity changes at each monitoring point, the changes in pollutant concentration are predicted.

[0023] Step 3: Iteratively adjust Q using a local model j The model uses the diffusion coefficient D to capture the nonlinear characteristics of pollutant diffusion with higher accuracy. For each iteration, the error E is recalculated and optimized. In the area near the pollution source predicted by the model, the pollutant concentration C at each monitoring point is recalculated. i This data is compared with the actual concentration value to further reduce errors. By minimizing local errors, the location of the pollution source is gradually approximated to a smaller range. Once the convergence reaches a lower level or meets the threshold, the water quality monitoring system will output the final pollution source location and specific source strength value Q. j ;

[0024] S7, Alarm Mechanism

[0025] All monitoring data and model analysis results are uploaded to the cloud platform in real time for comprehensive analysis. The water quality monitoring system assigns a risk level to each potential pollution source based on the accuracy of concentration prediction. The cloud platform combines rainfall data from the surrounding area and data collected by the equipment to further verify abnormal discharge situations. If it is determined that industrial wastewater is being illegally discharged during the flood season, the water quality monitoring system will automatically alarm and send the source tracing results to relevant regulatory departments or personnel to ensure timely response and handling.

[0026] Preferably, in step one, it is assumed that there are rainwater discharge outlets S1, S2, S3, S4, S5, and S6 in the river channel, located at coordinate S. j =(X j ,Y j (j=1,2,…,n), the equipment monitoring points are located at P1, P2, P3, P4, and at coordinate P. i =(X i ,Y i (i = 1, 2, ..., m), predict the S of each rainwater discharge outlet using the model. j For each monitoring point P i The effect of concentration;

[0027] Assuming the pollutants spread in a uniform diffusion manner, in order to assess the impact of each discharge outlet on the concentration at the monitoring points, the S value for each discharge outlet is first calculated. j To each monitoring point P i distance d ij :

[0028]

[0029] Using the pollutant diffusion coefficient D, the source strength Q predicted by the machine model, and the distance d ij Using the time variable t, the pollutant concentration C at the monitoring point is calculated using a diffusion model. i :

[0030]

[0031] At monitoring point P i The actual measured pollutant concentration was C. i The model predicts the pollutant concentration as C. i,obs Based on this, the sum of squared errors E is used to measure C. i and C i,obs Differences between them:

[0032]

[0033] Preferably, the formula used in step three is:

[0034]

[0035] Where η is the learning rate, used to control the step size of each iteration.

[0036] Preferably, the power supply module provides operating power to the network module, display control module, hydrological gateway module, sensor configuration module, and external sensors; the network module communicates with the private cloud platform via wired or wireless means; the hydrological sensor, disaster prevention and mitigation sensor, and agricultural drinking water sensor are connected to the hardware circuit board through sensor interfaces to collect data in real time and transmit it to the network module; the sensor configuration module is connected to the host computer software through an RS232 to USB module to realize sensor configuration and management; the hydrological gateway module uses serial port peripherals and AT commands for connection control and to send and receive data from the hydrological platform; the display control module displays the device status and sensor data.

[0037] Preferably, the power supply module is powered by an 8-36V DC voltage input.

[0038] Preferably, the sensor interface supports a variety of sensor interfaces and protocols, including but not limited to SDI-12, Gray code, RS485, RS232, pulse input, and analog-to-digital input / output types.

[0039] Preferably, the hydrological sensor is used to monitor hydrological parameters and calculate information on the rate of water level rise and changes in river water level; the disaster prevention and mitigation linkage sensor is used to display the collected indicator information and automatically trigger alarms based on water level thresholds and water level rise rate thresholds; the agricultural drinking water sensor is used to monitor water quality parameters, pipeline flow indicators, and electricity usage of agricultural drinking water, and realizes automatic control of valve opening or remote manual control.

[0040] Preferably, the network module includes a wired network interface and a wireless 4G module. The wired network uses a universal RJ45 network port, and the wireless 4G module uses a Nano-SIM card for installation, compatible with current mobile phone cards and IoT cards. Either type can be selected for installation according to the site conditions. If both are installed simultaneously, the wired network will be used first for data upload. The network module, based on the IoT-based flood season rainwater inlet sewage discharge monitoring device, packages the collected data according to rules and uploads it to the private cloud platform. At the same time, it receives and processes instructions and requests issued by the private cloud platform.

[0041] Preferably, the hydrological gateway module is connected to the hydrological management platform. The IoT-based flood season rainwater inlet sewage illegal discharge monitoring device collects data and uploads it to the hydrological management platform according to rules. At the same time, it receives and processes instructions and requests issued by the platform.

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] 1. This IoT-based monitoring device and early warning method for illegal sewage discharge from storm drains during the flood season, through the integration of IoT technology and a water quality monitoring system, achieves real-time monitoring and early warning of illegal industrial wastewater discharge from storm drains during the flood season. It solves the problems of high costs and poor real-time performance of manual monitoring, and the difficulty in tracing pollution locations due to the fact that online water quality monitoring equipment is mostly used for basic water quality parameter detection. It can efficiently identify water quality changes and accurately locate pollution sources, significantly improving the accuracy of sewage source tracing. Through intelligent and automated operation, it reduces reliance on manual monitoring and lowers labor costs.

[0044] 2. This IoT-based monitoring device for illegal sewage discharge from storm drains during the flood season and its early warning method, combined with big data analysis and machine learning algorithms, uses a dual-indicator judgment standard to identify sewage anomalies and a source tracing algorithm to locate the sewage discharge from storm drains. It can issue early warnings and trace the source in a timely manner, which can assist relevant departments in making more scientific and reasonable decisions, effectively protect the aquatic ecological environment, reduce the impact of pollution on water resources, and has significant environmental and social benefits. Attached Figure Description

[0045] Figure 1 This is a schematic diagram of the outer casing of the monitoring device for illegal sewage discharge during the flood season of this invention.

[0046] Figure 2 This is a block diagram of the overall water quality monitoring system operation of the rainwater inlet sewage illegal discharge monitoring equipment of the present invention during the flood season;

[0047] Figure 3 This is a schematic diagram of the power supply principle of the monitoring device for illegal sewage discharge during the flood season of the present invention.

[0048] Figure 4 This is a schematic diagram of the interface design of the monitoring device for illegal sewage discharge during the flood season of the present invention.

[0049] Figure 5 This is a flowchart of the monitoring of illegal sewage discharge from storm drains during the flood season according to the present invention;

[0050] Figure 6 This is a layout diagram of the river channel storm drain distribution and monitoring equipment of the present invention;

[0051] Figure 7 This is a rear view of the outer casing of the monitoring device for illegal sewage discharge during the flood season of this invention.

[0052] Figure 8 This is a schematic diagram of the left side of the outer casing of the monitoring device for illegal sewage discharge during the flood season of this invention. Detailed Implementation

[0053] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0054] refer to Figure 1 and Figure 2 This invention provides an IoT-based monitoring device for illegal sewage discharge from stormwater inlets during the flood season, comprising a waterproof casing, a hardware circuit board, a solar power supply system, a hydrological sensor, a disaster prevention and mitigation linkage sensor, an agricultural drinking water sensor, a private cloud platform, a hydrological platform, and host computer software; the hardware circuit board includes a power module, a network module, a display control module, a hydrological gateway module, a sensor configuration module, and a sensor interface.

[0055] See Figure 3 The power module provides operating power to the network module, display control module, hydrological gateway module, sensor configuration module, and external sensors. Considering that the equipment needs to operate around the rainwater inlet for extended periods, and to ensure normal operation while maximizing its service life, the power module supplies power to the entire equipment via an external 8-36V DC input power supply (lithium battery), and is charged using a 120W solar panel.

[0056] See Figure 4 The sensor interfaces cover various types, including SDI-12, Gray code, RS485, RS232, pulse input, and analog / digital input / output. By supporting multiple sensor interfaces and protocols, the IoT-based intelligent hydrological sensing device allows for program debugging via serial port. The hardware circuit board connects to hydrological sensors, disaster prevention and mitigation linkage sensors, and agricultural drinking water sensors through the sensor interfaces, enabling monitoring and control of these sensors.

[0057] Hydrological sensors include tipping bucket rain gauges, radar water level gauges, radar flow meters, float water level gauges, etc., which can monitor hydrological parameters such as rainfall, real-time water level, instantaneous flow velocity, instantaneous flow rate, and cumulative flow, and calculate information such as the rate of water level rise and changes in river water level.

[0058] Agricultural drinking water sensors include water quality analyzers, electromagnetic flow meters, smart meters, solenoid valves, and electric valves. They can monitor water quality parameters, pipeline flow rates, and electricity usage, and enable automatic control of valve opening or remote manual control.

[0059] Disaster prevention and mitigation linkage sensors include LED electronic displays, multi-color audible and visual alarms, and single-color audible and visual alarms. LED electronic displays can show collected hydrological, agricultural and drinking water indicators and provide early warnings. Multi-color and single-color audible and visual alarms can automatically trigger local alarms based on water level thresholds and water level rise rate thresholds.

[0060] The network module communicates with the private cloud platform via wired or wireless means. The network module includes a wired network interface and a wireless 4G module. The wired network uses a universal RJ45 network port, while the wireless 4G module uses a Nano-SIM card for installation, compatible with current mobile phone cards and IoT cards. Either method can be selected for installation depending on the site conditions; if both are installed simultaneously, the wired network will be used preferentially for data upload. Based on the IoT-enabled rainwater inlet sewage discharge monitoring device, the network module packages and uploads the collected data to the private cloud platform according to rules. Simultaneously, it receives and processes instructions and requests from the private cloud platform.

[0061] The sensor configuration module connects to the host computer software via an RS232 to USB converter to enable sensor configuration and management. The host computer software allows configuration of sensor types, parameters, and models, and can also read and modify the sensor configuration table within the IoT-based flood season stormwater inlet sewage discharge monitoring device, enabling flexible access and configuration of multiple sensors.

[0062] The hydrological gateway module uses serial port peripherals and AT commands for connection and control, and sends and receives data from the hydrological platform. The hydrological gateway module is connected to the hydrological management platform. The IoT-based flood season rainwater inlet sewage discharge monitoring device packages the collected data according to rules and uploads it to the hydrological management platform. At the same time, it receives and processes the instructions and requests issued by the platform.

[0063] The display control module shows the device status and sensor data. It includes a 2.4-inch 128*64 resolution LCD screen, four touch buttons, and four LED status indicators. The display provides rich and comprehensive information with clear colors. The entire LCD panel is replaceable, saving on subsequent maintenance costs and time. The four touch buttons allow users to flip through pages of information on the display, and the four LED status indicators show the current operating status of the device.

[0064] This invention provides an early warning method for monitoring illegal sewage discharge at stormwater inlets during the flood season based on the Internet of Things (IoT). The method first installs monitoring equipment at river stormwater discharge outlets to collect historical rainfall, water level, and water quality data from the stormwater inlets during the flood season. This data is then used to build a machine learning model. Real-time collected monitoring data is input into the trained model to predict water conductivity. Based on a dual-standard judgment method using conductivity values, if industrial wastewater is identified, a water quality source tracing model is immediately activated to track the pollution source and report the data to a cloud platform. A flowchart of the monitoring event for illegal sewage discharge at stormwater inlets during the flood season is attached. Figure 5 As shown.

[0065] The early warning method includes the following steps:

[0066] S1. Model Establishment

[0067] Historical rainfall, water level, and water quality data from flood seasons are used to train machine learning models. The model inputs include rainfall, water level change rate, and pH value, and the output is conductivity. Using this historical data, machine learning models such as XGBoost, LSTM, and Random Forest are built.

[0068] S2, Data Acquisition

[0069] The monitoring equipment is installed around the rainwater outfall to collect rainfall, water level, and water quality data in real time and report them to the cloud platform. The hydrological sensors inside the monitoring equipment record the cumulative rainfall CP(t) and the periodic rainfall P(t), and the cumulative rainfall is periodically reset to prevent data overflow. They also record the water level Z(t) and are periodically calibrated to ensure measurement accuracy. The agricultural drinking water sensors measure pH value and conductivity.

[0070] S3, Dynamic Data Processing

[0071] The water quality monitoring system within the cloud platform calculates the rainfall rate based on real-time collected monitoring data during the operation of the monitoring equipment. and rate of change of water level At the same time, the monitoring equipment will monitor the conductivity k of the water in real time through agricultural drinking water sensors. obs The standard deviation σ of the electrical conductivity of rainwater was calculated based on the time series.

[0072] S4, Model Input

[0073] Data such as cumulative rainfall, rainfall over a period of time, rainfall rate, water level and its rate of change, and pH value are input into a pre-trained machine learning model in real time. Based on the patterns learned during training, the model outputs the predicted conductivity of the water body. And predict the source strength Q;

[0074] S5. Anomaly Detection

[0075] The system employs risk-based tiered monitoring and adaptive sampling modes. Risk-based monitoring means that when the water conductivity is normal, the water quality monitoring system maintains a basic sampling frequency of once per hour; when the conductivity becomes abnormal, it automatically increases to a sampling frequency of once every 15 minutes. Adaptive sampling adjustment means that if the monitoring data changes significantly, even if no warning value is triggered, it will switch to high-frequency sampling, i.e., once per minute, and upload the data to the cloud platform. When the sensor detects high-intensity rainfall, the water quality monitoring system automatically activates the high-frequency sampling mode, and resumes low-frequency sampling after the data stabilizes, thereby optimizing energy efficiency.

[0076] A dual-indicator judgment standard is adopted: if Consider it normal rainwater; if and Consider it a minor abnormality, record and continue to observe, maintaining the baseline sampling frequency of one hour; if and Entering monitoring mode, the sampling frequency is increased to every 15 minutes; if and Treating this as a special case, and combining actual analysis and subsequent monitoring, assess whether there is a risk of illegal sewage discharge; if and The high-frequency sampling mode is activated, which means sampling once per minute and reporting to the water quality monitoring system to mark the possible illegal discharge of sewage, thus entering the source tracing analysis stage;

[0077] S6, Source Tracing Analysis

[0078] When the water quality monitoring system detects excessive conductivity, it immediately activates the water quality source tracing model. This model uses reverse inference analysis, combining water quality monitoring data from multiple known upstream stormwater inlets with real-time concentration data from equipment monitoring points, to locate the pollution source. The water quality source tracing model employs a multi-level source tracing algorithm, using deep learning combined with traditional diffusion models to gradually narrow down the pollution source range and accurately pinpoint its location.

[0079] The specific algorithm is as follows:

[0080] Step 1: Initially locate the pollution source range using an approximate optimization algorithm, and adjust the source strength Q using an optimization algorithm. j To reduce the sum of squared errors E, thereby identifying potential sewage discharge areas;

[0081] Assume there are rainwater and sewage outlets S1, S2, S3, S4, S5, and S6 in the river channel, located at coordinate S. j =(X j ,Y j (j=1,2,…,n), the equipment monitoring points are located at P1, P2, P3, P4, and at coordinate P. i =(X i ,Y i(i = 1, 2, ..., m), predict the S of each rainwater discharge outlet using the model. j For each monitoring point P i The concentration effect, the distribution of river storm drains and the layout of monitoring equipment are shown in the attached diagram. Figure 6 As shown;

[0082] Assuming the pollutants spread in a uniform diffusion manner, in order to assess the impact of each discharge outlet on the concentration at the monitoring points, the S value for each discharge outlet is first calculated. j To each monitoring point P i distance d ij :

[0083]

[0084] Using the pollutant diffusion coefficient D, the source strength Q predicted by the machine model, and the distance d ij Using the time variable t, the pollutant concentration C at the monitoring point is calculated using a diffusion model. i :

[0085]

[0086] At monitoring point P i The actual measured pollutant concentration was C. i The model predicts the pollutant concentration as C. i,obs Based on this, the sum of squared errors E is used to measure C. i and C i,obs Differences between them:

[0087]

[0088] Step 2: In the initially identified potential pollution source areas, use a deep learning model to conduct a more detailed concentration distribution analysis. The input to the deep learning model includes Q obtained from the preliminary source tracing analysis. j Real-time monitoring data: rainfall, water level, conductivity, pH value. Based on the relative coordinates between each monitoring point and the pollution source, as well as the propagation path of pollutants in the water body, and the time series data of pollutant concentration changes over time, including the conductivity changes at each monitoring point, the changes in pollutant concentration are predicted.

[0089] Step 3: Iteratively adjust Q using a local model j And the diffusion coefficient D, to capture the nonlinear characteristics of pollutant diffusion with higher accuracy, and for each iteration, the error E is recalculated and optimized:

[0090]

[0091] Where: η is the learning rate, used to control the step size of each iteration;

[0092] In the area near the pollution source predicted by the model, the pollutant concentration C at each monitoring point was recalculated. i And compare it with the actual concentration value to further reduce the error:

[0093]

[0094] By minimizing local errors, the location of the pollution source is gradually approximated to a smaller range. Once the convergence reaches a lower level or meets the threshold, the water quality monitoring system will output the final pollution source location and specific source strength value Q. j ;

[0095] Water quality source tracing models can adopt current analytical models and statistical analysis models. Approximate optimization algorithms can be particle swarm optimization or simulated backoff algorithms. Deep learning models can be a combination of LSTM and CNN or a combination of LSTM and GCN. Optimization algorithms can be least squares and genetic algorithms.

[0096] S7, Alarm Mechanism

[0097] All monitoring data and model analysis results are uploaded to the cloud platform in real time for comprehensive analysis. The water quality monitoring system assigns a risk level to each potential pollution source based on the accuracy of concentration prediction. The cloud platform combines rainfall data from the surrounding area and data collected by the equipment to further verify abnormal discharge situations. If it is determined that industrial wastewater is being illegally discharged during the flood season, the water quality monitoring system will automatically alarm and send the source tracing results to relevant regulatory departments or personnel to ensure timely response and handling.

[0098] As described above, this IoT-based early warning method for monitoring illegal sewage discharge from storm drains during the flood season combines IoT technology with a water quality monitoring system. It achieves real-time monitoring and early warning of illegal industrial wastewater discharge from storm drains during the flood season, solving the problems of high costs and poor real-time performance associated with manual monitoring. Furthermore, online water quality monitoring equipment is primarily used for basic water quality parameter detection, making it difficult to trace the location of pollution. By combining big data analysis and machine learning algorithms, a dual-indicator judgment standard is used to identify abnormal sewage discharge, and a source tracing algorithm is used to locate the discharging storm drain. This reduces reliance on manual monitoring and lowers labor costs. Furthermore, by combining big data analysis and machine learning algorithms, this invention can assist relevant departments in making more scientific and rational decisions. These functions effectively protect the aquatic ecological environment, reduce the impact of pollution on water resources, and have significant environmental and social benefits.

[0099] All standard parts used in this invention can be purchased from the market, and irregularly shaped parts can be customized according to the description and drawings. The specific connection methods of each part all adopt conventional means such as bolts that are mature in the prior art. The machinery, parts and equipment all adopt conventional models in the prior art, which will not be described in detail here. The contents not described in detail in this specification belong to the prior art known to those skilled in the art. Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An early warning method for monitoring sewage discharge from storm drains during the flood season based on the Internet of Things, characterized in that: This early warning method first installs monitoring equipment at river stormwater outfalls. This equipment collects historical data on rainfall, water level, and water quality at the outfalls during the flood season. This data is then used to build a machine learning model. Real-time monitoring data is input into the trained model to predict water conductivity. Based on a dual-standard judgment method using conductivity values, if the water is industrial wastewater, a water quality source tracing model is immediately activated to track the pollution source and report it to a cloud platform. The monitoring equipment includes a waterproof casing, a hardware circuit board, a solar power system, hydrological sensors, disaster prevention and mitigation linkage sensors, agricultural and drinking water sensors, a private cloud platform, a hydrological platform, and host computer software. The hardware circuit board includes a power module, a network module, a display and control module, a hydrological gateway module, a sensor configuration module, and sensor interfaces. The early warning method includes the following steps: S1. Model Establishment Historical rainfall, water level, and water quality data from the flood season are used to train a machine learning model. The model input includes rainfall, water level change rate, and pH value, and the output is conductivity. This historical data is used to build a machine learning model. S2, Data Acquisition The monitoring equipment is installed around the rainwater outfall to collect rainfall, water level, and water quality data in real time and report them to the cloud platform. The hydrological sensors inside the monitoring equipment record the cumulative rainfall CP(t) and the periodic rainfall P(t), and the cumulative rainfall is periodically reset to prevent data overflow. They also record the water level Z(t) and are periodically calibrated to ensure measurement accuracy. The agricultural drinking water sensors measure pH value and conductivity. S3, Dynamic Data Processing The water quality monitoring system within the cloud platform calculates the rainfall rate based on real-time collected monitoring data during the operation of the monitoring equipment. and rate of change of water level At the same time, the monitoring equipment will monitor the conductivity k of the water body in real time through agricultural drinking water sensors. obs The standard deviation σ of the electrical conductivity of rainwater was calculated based on the time series. S4, Model Input The cumulative rainfall, rainfall over a period of time, rainfall rate, water level and its rate of change, and pH value data are input into a pre-trained machine learning model in real time. Based on the patterns learned during training, the model outputs the predicted conductivity of the water body. And predict the source strength Q; S5. Anomaly Detection The system employs risk-based tiered monitoring and adaptive sampling modes. Risk-based monitoring means that when the water conductivity is normal, the water quality monitoring system maintains a basic sampling frequency of once per hour; when the conductivity becomes abnormal, it automatically increases to a sampling frequency of once every 15 minutes. Adaptive sampling adjustment means that if the monitoring data changes significantly, even if no warning value is triggered, it will switch to high-frequency sampling, i.e., once per minute, and upload the data to the cloud platform. When the sensor detects high-intensity rainfall, the water quality monitoring system automatically activates the high-frequency sampling mode, and resumes low-frequency sampling after the data stabilizes, thereby optimizing energy efficiency. A dual-indicator judgment standard is adopted: if Consider it normal rainwater; if and Consider it a minor abnormality, record and continue to observe, maintaining the baseline sampling frequency of one hour; if and Entering monitoring mode, the sampling frequency is increased to every 15 minutes; if and Treating this as a special case, and combining actual analysis and subsequent monitoring, assess whether there is a risk of illegal sewage discharge; if and The high-frequency sampling mode is activated, which means sampling once per minute and reporting to the water quality monitoring system to mark the possible illegal discharge of sewage, thus entering the source tracing analysis stage; S6, Source Tracing Analysis When the water quality monitoring system detects that the conductivity exceeds the standard, the water quality source tracing model is immediately activated. The water quality source tracing model uses reverse inference analysis, combined with water quality monitoring data from multiple known upstream rainwater inlets and real-time concentration data from equipment monitoring points, to locate the pollution source. The water quality source tracing model adopts a multi-level source tracing algorithm, which uses deep learning combined with traditional diffusion models to gradually narrow down the range of pollution sources and accurately locate them. The specific algorithm is as follows: Step 1: Initially locate the pollution source range using an approximate optimization algorithm, and adjust the source strength Q using an optimization algorithm. j To reduce the sum of squared errors E, thereby identifying potential sewage discharge areas; Step 2: In the initially identified potential pollution source areas, use a deep learning model to conduct a more detailed concentration distribution analysis. The input to the deep learning model includes Q obtained from the preliminary source tracing analysis. j Real-time monitoring data: rainfall, water level, conductivity, pH value. Based on the relative coordinates between each monitoring point and the pollution source, as well as the propagation path of pollutants in the water body, and the time series data of pollutant concentration changes over time, including the conductivity changes at each monitoring point, the changes in pollutant concentration are predicted. Step 3: Iteratively adjust Q using a local model j The model uses the diffusion coefficient D to capture the nonlinear characteristics of pollutant diffusion with higher accuracy. For each iteration, the error E is recalculated and optimized. In the area near the pollution source predicted by the model, the pollutant concentration C at each monitoring point is recalculated. i This data is compared with the actual concentration value to further reduce errors. By minimizing local errors, the location of the pollution source is gradually approximated to a smaller range. Once the convergence reaches a lower level or meets the threshold, the water quality monitoring system will output the final pollution source location and specific source strength value Q. j ; S7, Alarm Mechanism All monitoring data and model analysis results are uploaded to the cloud platform in real time for comprehensive analysis. The water quality monitoring system assigns a risk level to each potential pollution source based on the accuracy of concentration prediction. The cloud platform combines rainfall data from the surrounding area and data collected by the equipment to further verify abnormal discharge situations. If it is determined that industrial wastewater is being illegally discharged during the flood season, the water quality monitoring system will automatically alarm and send the source tracing results to relevant regulatory departments or personnel to ensure timely response and handling.

2. The early warning method for the monitoring equipment for illegal sewage discharge during the flood season based on the Internet of Things as described in claim 1, characterized in that: In step one, assume there are rainwater discharge outlets S1, S2, S3, S4, S5, and S6 in the river channel, located at coordinate S. j =(X j ,Y j (j=1,2,…,n), the equipment monitoring points are located at P1, P2, P3, P4, and are located at coordinate P. i =(X i ,Y i (i = 1, 2, ..., m), predict the S of each rainwater discharge outlet using the model. j For each monitoring point P i The effect of concentration; Assuming the pollutants spread in a uniform diffusion manner, in order to assess the impact of each discharge outlet on the concentration at the monitoring points, the S value for each discharge outlet is first calculated. j To each monitoring point P i distance d ij : Using the pollutant diffusion coefficient D, the source strength Q predicted by the machine model, and the distance d ij Using the time variable t, the pollutant concentration C at the monitoring point is calculated using a diffusion model. i : At monitoring point P i The actual measured pollutant concentration was C. i The model predicts the pollutant concentration as C. i,obs Based on this, the sum of squared errors E is used to measure C. i and C i,obs Differences between them:

3. The early warning method for the monitoring equipment for illegal sewage discharge during the flood season based on the Internet of Things as described in claim 2, characterized in that: The formula used in step three is: Where η is the learning rate, used to control the step size of each iteration.

4. The early warning method for the monitoring equipment for illegal sewage discharge from rainwater inlets during the flood season based on the Internet of Things, as described in claim 3, is characterized in that: The power module provides operating power to the network module, display control module, hydrological gateway module, sensor configuration module, and external sensors. The network module communicates with the private cloud platform via wired or wireless means. The hydrological sensors, disaster prevention and mitigation sensors, and agricultural drinking water sensors are connected to the hardware circuit board through sensor interfaces to collect data in real time and transmit it to the network module. The sensor configuration module connects to the host computer software via an RS232 to USB module to configure and manage the sensors. The hydrological gateway module uses serial port peripherals and AT commands for connection control, sending and receiving data from the hydrological platform. The display control module displays the device status and sensor data.

5. The early warning method for the monitoring equipment for illegal sewage discharge from rainwater inlets during the flood season based on the Internet of Things, as described in claim 4, is characterized in that: The power supply module is powered by an 8-36V DC voltage input.

6. The early warning method for the monitoring equipment for illegal sewage discharge from rainwater inlets during the flood season based on the Internet of Things, as described in claim 4, is characterized in that: The sensor interface supports a variety of sensor interfaces and protocols, including but not limited to SDI-12, Gray code, RS485, RS232, pulse input, and analog-to-digital input / output types.

7. The early warning method for the monitoring equipment for illegal sewage discharge from rainwater inlets during the flood season based on the Internet of Things, as described in claim 4, is characterized in that: The hydrological sensor is used to monitor hydrological parameters and calculate information on water level rise rate and river water level changes; the disaster prevention and mitigation linkage sensor is used to display the collected index information and automatically trigger alarms based on water level thresholds and water level rise rate thresholds; the agricultural drinking water sensor is used to monitor water quality parameters, pipeline flow rate, and electricity usage of agricultural drinking water, and realizes automatic control of valve opening or remote manual control.

8. The early warning method for the monitoring equipment for illegal sewage discharge from rainwater inlets during the flood season based on the Internet of Things, as described in claim 4, is characterized in that: The network module includes a wired network interface and a wireless 4G module. The wired network uses a universal RJ45 network port, and the wireless 4G module uses a Nano-SIM card for installation, compatible with current mobile phone cards and IoT cards. Either type can be selected for installation according to the site conditions. If both are installed simultaneously, the wired network will be used first for data upload. The network module, based on the IoT-based flood season rainwater inlet sewage discharge monitoring device, packages the collected data according to rules and uploads it to the private cloud platform. At the same time, it receives and processes instructions and requests issued by the private cloud platform.

9. The early warning method for the monitoring equipment for illegal sewage discharge from rainwater inlets during the flood season based on the Internet of Things, as described in claim 4, is characterized in that: The hydrological gateway module is connected to the hydrological management platform. The IoT-based monitoring device for illegal sewage discharge during the flood season packages the collected data and uploads it to the hydrological management platform according to the rules. At the same time, it receives and processes the instructions and requests issued by the platform.

Images