Intelligent water quality monitoring system based on multi-source data fusion

Abstract

The invention discloses an intelligent water quality monitoring system based on multi-source data fusion, and relates to the technical field of water quality monitoring, the intelligent water quality monitoring system comprises a ground management and control platform and a monitoring device, the ground management and control platform is used for regularly collecting and identifying original data, and processing, analyzing and early warning the original data; the monitoring device is used for acquiring multi-source original data related to a water quality monitoring area and performing all-directional monitoring on scene information and environmental data of the water quality monitoring area, and the monitoring device comprises a water quality monitoring station, a satellite remote sensing device, a mobile monitoring device and a meteorological station. By integrating a multi-source data acquisition mode, water quality information can be captured in all directions, the capacity of capturing water quality changes of a complex drainage basin is improved, the defect of a single data source is effectively overcome, all the monitoring devices collect data in real time and transmit the data to the ground management and control platform, and the data is analyzed through a data processing and analysis process. And a water quality monitoring result and a prediction trend can be output in real time, so that managers can timely perceive water quality changes.

Description

Technical Field

[0001] This invention relates to the field of water quality monitoring technology, specifically to an intelligent water quality monitoring system based on multi-source data fusion. Background Technology

[0002] With the acceleration of industrialization and urbanization, water pollution has become increasingly severe, placing higher demands on water quality monitoring. Traditional water quality monitoring methods often rely on single monitoring technologies, such as ground-based water quality monitoring stations, which can only obtain water quality data for a limited area around the monitoring station. This lacks a macroscopic understanding of the overall water quality of large water areas. While satellite remote sensing technology can achieve large-scale coverage, it lacks precision in monitoring micro-indicators of water quality and is easily affected by cloud cover and atmospheric interference, which can impact data accuracy. Furthermore, environmental factors such as meteorological conditions and watershed topography can affect water quality change monitoring. However, most existing water quality monitoring systems fail to fully consider the fusion of multi-source data, making it difficult to comprehensively and accurately reflect the true state and changing trends of water quality.

[0003] Based on this, we now offer an intelligent water quality monitoring system based on multi-source data fusion, which can eliminate the drawbacks of existing technical solutions. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent water quality monitoring system based on multi-source data fusion, so as to solve the problem that traditional water quality monitoring methods in the background art mostly rely on a single monitoring technology and do not consider multi-source data fusion.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] The intelligent water quality monitoring system based on multi-source data fusion includes a ground control platform and monitoring devices that communicate with each other via a network. The ground control platform is used to periodically collect and identify raw data from different sources and different water quality monitoring areas, and to process, analyze, and issue early warnings for the raw data. The monitoring devices are used to acquire multi-source raw data related to the water quality monitoring area, and to conduct comprehensive monitoring of the scene information and environmental data of the water quality monitoring area. The monitoring devices include water quality monitoring stations, satellite remote sensing, mobile monitoring equipment, and weather stations. The water quality monitoring stations, satellite remote sensing, mobile monitoring equipment, and weather stations all transmit the collected raw data to the ground control platform via a wireless communication network. The ground control platform includes:

[0007] The data preprocessing module is used to preprocess the collected multi-source raw data, perform normalization operations and remove duplicate and abnormal data, add timestamps to make the water quality data from different data sources consistent in terms of presentation, units and format, and build the initial dataset.

[0008] The feature extraction module is used to analyze the monitoring data in the initial dataset and extract key feature data;

[0009] The weight adjustment module is used to adjust the weights of each data source in real time based on meteorological data, parameters of the monitoring device's movement process, and water quality indicators.

[0010] The data fusion module is used to fuse multi-source water quality data, integrating data from different sources and of different types to generate a fused dataset;

[0011] The model building module is used to build a water quality monitoring model based on the fused dataset and key feature data, and to monitor and output the operation status and prediction results of the water quality monitoring model in real time.

[0012] The monitoring and evaluation module is used to output the operation status and prediction results of the water quality monitoring model in a visual form, so as to facilitate the timely detection of water quality anomalies.

[0013] Preferably, the water quality monitoring station is equipped with various types of sensors to collect real-time water quality scene information and environmental data for the corresponding water quality monitoring area, specifically including:

[0014] The first data acquisition unit is equipped with a pH sensor, dissolved oxygen sensor, turbidity sensor, COD sensor, ammonia nitrogen sensor, detector, and high-precision camera, which are used to collect water quality indicators and watershed parameters in the corresponding water quality monitoring area in real time.

[0015] The first data transmission unit is responsible for aggregating the raw data collected by several sensors and transmitting it to the ground control platform via a wireless communication network.

[0016] The first GPS positioning unit is used to obtain the geographical coordinates of the water quality monitoring station and associate and bind the positioning information with the raw data collected by the first data acquisition unit.

[0017] Preferably, the satellite remote sensing is used to acquire information on water body distribution and water level changes in a large-scale water quality monitoring area, specifically including:

[0018] The second data acquisition unit is used to collect electromagnetic wave reflection and radiation signals of the water quality monitoring area through optical sensors and radar sensor equipment carried by the satellite. The signals include the spectral and morphological characteristics of the water body and the surrounding materials.

[0019] The second data transmission unit is used to transmit the raw data acquired by the second data acquisition unit to the ground control platform via a satellite communication link.

[0020] The second GPS positioning unit is used to record the satellite orbit parameters, observation time, and geographical coordinates of the corresponding ground monitoring area during remote sensing data acquisition using the Global Positioning System, and to associate and bind the positioning information with the raw data acquired by the second data acquisition unit.

[0021] Preferably, the mobile monitoring equipment includes unmanned aerial vehicle (UAV) equipment and unmanned surface vessel (USV) equipment, used for real-time on-site water quality measurement operations, specifically including:

[0022] The third data acquisition unit includes a UAV equipped with a multispectral camera and a gas sensor for acquiring images of the water area and volatile organic compound concentration data, and an unmanned boat equipped with a water quality sensor for acquiring water quality indicators and watershed parameters from multiple water samples.

[0023] The third data transmission unit is used to transmit the raw data collected by the third data acquisition unit to the ground control platform in real time via a wireless communication network.

[0024] The third GPS positioning unit is used to obtain the real-time position coordinates of the UAV during flight, as well as the specific coordinates of the UAV when it moves to collect water samples in the water quality monitoring area, and to associate and bind the positioning information with the corresponding collected data.

[0025] Preferably, the meteorological station is used to acquire meteorological data related to hydrology and water quality, specifically including:

[0026] The fourth data acquisition unit is used to acquire meteorological data related to hydrology and water quality. It is equipped with rain gauges, temperature and humidity sensors, wind speed sensors, visibility meters, and sunshine meters to collect meteorological data related to hydrology and water quality in real time.

[0027] The fourth data transmission unit is used to transmit the meteorological data acquired by the fourth data acquisition unit to the ground control platform via a wireless communication network.

[0028] The fourth GPS positioning unit is used to obtain the geographical coordinates of the meteorological station and associate the positioning information with the meteorological data collected by the fourth data acquisition unit.

[0029] Preferably, the process of the feature extraction module extracting key feature data specifically includes:

[0030] Monitoring data is extracted from the initial dataset, the correlation of variables in the initial dataset is analyzed, and key feature data with high correlation are identified. The key feature data includes collection source information, time and location, water quality indicators, watershed parameters and meteorological data.

[0031] For feature extraction of the collected source information, the unique identifier of the data source is determined by recording the equipment number, type and monitoring network node information of each monitoring device, and the stability parameters of the data transmission link are also recorded.

[0032] For the extraction of time and location features, the latitude and longitude coordinates of each monitoring point are obtained through the GPS positioning unit built into the monitoring device. Combined with the electronic map data of the water quality monitoring area, the latitude and longitude coordinates are converted into watershed partition codes. The data collection time is recorded using timestamps, and the time dimension features including year, month, day, hour and minute are decomposed and marked as whether it is a peak period for water quality monitoring.

[0033] For the feature extraction of water quality indicators, pH value is obtained by measuring the hydrogen ion concentration in the water body using a pH sensor, dissolved oxygen content in the water body is detected by a dissolved oxygen sensor, spectral features related to algae are extracted from the spectral data of remote sensing data to reflect the density and species distribution of algae in the water body, and algae species and density are identified by capturing images and video data with a high-precision camera and using edge detection and texture analysis technology.

[0034] For the feature extraction of watershed parameters, the water level in the water body is measured by a water level gauge, the flow velocity in the water quality monitoring area is measured by a flow meter, the particle size distribution and concentration of suspended particulate matter in the water body are measured by a laser particle size analyzer, the turbidity value of the water body is obtained by a turbidity sensor, and pollution indicators including ammonia nitrogen concentration, total phosphorus and total nitrogen are detected by a water quality analyzer. Geographic feature parameters including watershed slope, vegetation coverage and water area are extracted by combining topographic mapping data.

[0035] For feature extraction of meteorological data, precipitation is collected by rain gauges and converted into electrical signals to record rainfall; temperature and humidity sensors are used to record the temperature and relative humidity of the surrounding environment; wind speed is calculated using wind speed sensors; sunshine duration and intensity are recorded by sunshine meters; and weather visibility is measured by visibility meters.

[0036] All the key feature data extracted above were standardized, and the correlation between features was analyzed using Pearson correlation coefficient to identify highly correlated feature combinations. PCA dimensionality reduction technique was applied to reduce the number of features and retain principal components.

[0037] Preferably, the weight adjustment module adopts a hierarchical weight calculation method. The hierarchical weight includes individual weight and multi-device collaborative weight. The individual weight is used to adjust the weight of each monitoring parameter within a single type of monitoring device according to the differences in the monitoring focus of its monitoring points under different weather conditions, environments, and watersheds. The multi-device collaborative weight is used to adjust the parameter weight ratio of each monitoring device in the model construction according to the differences in the monitoring parameters of different monitoring devices when all monitoring devices are monitoring the same target object.

[0038] Preferably, the formula for calculating the individual weight is:

[0039] ;

[0040] in, For the first The first type of monitoring device Individual weights of each monitoring parameter, , , and All are weighted parameters. It can be adapted and adjusted according to actual needs. For the first The first type of monitoring device The basic weights of each monitoring parameter This is a function of the meteorological influence coefficient. For environmental impact coefficient function, This is a function of the location influence coefficient;

[0041] The formula for calculating the multi-device collaborative weight is as follows:

[0042] ;

[0043] in, For the first Multi-device collaborative weighting of similar monitoring devices For the first The monitoring coverage of this type of monitoring device For the first The monitoring accuracy of this type of monitoring device The number of monitoring device types participating in collaborative monitoring. For all Monitoring devices that participate in collaborative monitoring Perform a summation operation. .

[0044] Preferably, the process of constructing the water quality monitoring model in the model construction module specifically includes:

[0045] Based on the monitoring data in the initial dataset and the fused dataset, key feature data are extracted, and a feature dataset containing the extracted key feature data is constructed. The feature dataset is divided into a training set and a test set. The feature dataset includes the associated data of the collection source information, time and location, water quality indicators, watershed parameters and meteorological data.

[0046] Based on the needs of water quality monitoring, and combined with machine learning algorithms suitable for multi-source data, a water quality monitoring model is constructed using the training set. During the model construction process, the individual weights and multi-device collaborative weights obtained by the weight adjustment module are incorporated to enhance the reasonable contribution of different data sources to the water quality monitoring results.

[0047] The monitoring accuracy of the water quality monitoring model was evaluated using a test set, and different performance indicators were calculated and compared.

[0048] The differences between the monitoring results of the water quality monitoring model and the actual water quality are analyzed to identify potential sources of error. The model parameters are then adjusted, and the trained water quality monitoring model is deployed to the production environment of the ground control platform to receive updated multi-source fusion data in real time, as well as the operation status and prediction results of the water quality monitoring model.

[0049] Preferably, the water quality indicators include pH value, dissolved oxygen, algae density and total bacterial count; the watershed parameters include water level, flow velocity, suspended particulate matter concentration, turbidity, pollution level and geographical features; and the meteorological data include rainfall, temperature and humidity, wind speed, solar radiation intensity and weather visibility.

[0050] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0051] This invention establishes an intelligent water quality monitoring system based on multi-source data fusion. By integrating multiple data acquisition methods such as water quality monitoring stations, satellite remote sensing, mobile monitoring equipment, and meteorological stations, it can comprehensively capture water quality information, eliminate monitoring blind spots, and improve the ability to capture water quality changes in complex watersheds. Multi-source data fusion effectively compensates for the shortcomings of single data sources, comprehensively improving monitoring accuracy. Each monitoring device collects data in real time and transmits it quickly to the ground control platform through a data transmission unit. After data processing and analysis, it can output water quality monitoring results and predicted trends in real time, enabling managers to promptly detect water quality changes. In addition, the system can monitor external environmental parameters in real time, as well as environmental, time, weather, location, and water conditions during the movement of monitoring devices to corresponding locations, and the indicators parameters involved in water quality. The weight adjustment module dynamically adjusts data weights according to different monitoring scenarios, and the data fusion module generates a fused dataset. The water quality monitoring model built on this basis can summarize data to assess water quality conditions, facilitating timely detection of anomalies and the development of targeted measures. Attached Figure Description

[0052] Figure 1 This is a schematic diagram of the intelligent water quality monitoring system of the present invention.

[0053] Figure 2 This is a schematic diagram of the ground control platform of the present invention.

[0054] Figure 3This is a schematic diagram of the overall structure of the monitoring device of the present invention.

[0055] Figure 4 This is a schematic diagram of the unit structure of each device in the monitoring device of the present invention.

[0056] Figure label annotations: Ground control platform 100, data preprocessing module 110, feature extraction module 120, weight adjustment module 130, data fusion module 140, model building module 150, monitoring and evaluation module 160, water quality monitoring station 200, first data acquisition unit 210, first data transmission unit 220, first GPS positioning unit 230, satellite remote sensing 300, second data acquisition unit 310, second data transmission unit 320, second GPS positioning unit 330, mobile monitoring equipment 400, third data acquisition unit 410, third data transmission unit 420, third GPS positioning unit 430, meteorological station 500, fourth data acquisition unit 510, fourth data transmission unit 520, fourth GPS positioning unit 530. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0058] Example 1

[0059] In this embodiment, as Figure 1 - Figure 4 As shown, the intelligent water quality monitoring system based on multi-source data fusion includes a ground control platform 100 and monitoring devices that communicate with each other via a network. The ground control platform 100 is used to periodically collect and identify raw data from different sources and different water quality monitoring areas, and to process, analyze, and issue early warnings for the raw data. The monitoring devices are used to acquire multi-source raw data related to the water quality monitoring area, and to conduct comprehensive monitoring of the scene information and environmental data of the water quality monitoring area. The monitoring devices include a water quality monitoring station 200, a satellite remote sensing 300, a mobile monitoring device 400, and a meteorological station 500. The water quality monitoring station 200 can continuously monitor at a fixed location, the satellite remote sensing 300 is not limited by the ground environment, the mobile monitoring device 400 can be flexibly deployed, and the meteorological station 500 provides environmental data in real time. The above monitoring devices ensure that water quality-related information can be stably and accurately acquired in various complex environments. The water quality monitoring station 200, the satellite remote sensing 300, the mobile monitoring device 400, and the meteorological station 500 all transmit the collected raw data to the ground control platform 100 through a wireless communication network. The ground control platform 100 includes:

[0060] The data preprocessing module 110 is used to preprocess the collected multi-source raw data, perform normalization operations and remove duplicate and abnormal data, add timestamps, so that the water quality data from different data sources are consistent in terms of presentation, units and format, and construct the initial dataset.

[0061] Feature extraction module 120 is used to analyze the monitoring data in the initial dataset and extract key feature data;

[0062] The weight adjustment module 130 is used to adjust the weights of each data source in real time based on meteorological data, parameters of the monitoring device movement process, and water quality indicators.

[0063] Specifically, weight adjustment is to ensure that monitoring data from different sources and of different types are assigned a reasonable degree of importance based on their reliability, accuracy, applicability, and other factors during the subsequent model construction process. For example, in a certain water quality monitoring area, the data from satellite remote sensing 300 is greatly affected by cloud cover, resulting in reduced reliability, and thus its weight needs to be lowered. On the other hand, the monitoring data from water quality monitoring station 200 in this area is stable and accurate over a long period of time, and thus its weight needs to be increased.

[0064] The data fusion module 140 is used to fuse multi-source water quality data and integrate data from different sources and of different types to generate a fused dataset.

[0065] Specifically, the data fusion module 140 is used to perform fusion processing on multi-source water quality data based on the weight values ​​output by the weight adjustment module 130, integrating data from different sources and of different types to generate a fused dataset. During the fusion process, data integration is achieved through weighted summation, and the fused data values... ,in, To merge data values, For the first Preprocessed data collected by monitoring devices For the first Collaborative weights of monitoring devices;

[0066] The model building module 150 is used to build a water quality monitoring model based on the fused dataset and key feature data, and to monitor and output the operation status and prediction results of the water quality monitoring model in real time.

[0067] Specifically, by adjusting the weights to ensure that the multi-source data has a reasonable weight allocation before fusion, and then building a model based on this higher-quality fused data, the model can more accurately reflect the water quality and improve the model's performance and reliability.

[0068] The monitoring and evaluation module 160 is used to output the operation status and prediction results of the water quality monitoring model in a visual form, so as to facilitate the timely detection of water quality anomalies.

[0069] Among them, such as Figure 3 and Figure 4 As shown, the water quality monitoring station 200 is equipped with various types of sensors to collect real-time water quality scene information and environmental data for the corresponding water quality monitoring area, specifically including:

[0070] The first data acquisition unit 210 is equipped with a pH sensor, a dissolved oxygen sensor, a turbidity sensor, a COD sensor, an ammonia nitrogen sensor, a detector, and a high-precision camera. The first data acquisition unit 210 also includes other sensors for real-time acquisition of water quality indicators and watershed parameters of the corresponding water quality monitoring area, real-time acquisition of water quality scene information of the corresponding water quality monitoring area such as water color and turbidity, as well as environmental data such as pH value, dissolved oxygen content, chemical oxygen demand, ammonia nitrogen concentration, and other water quality indicators in the water body, and conversion of physical and chemical signals into processable raw data.

[0071] The first data transmission unit 220 is responsible for summarizing the raw data collected by several sensors and transmitting it to the ground control platform 100 through a wireless communication network. The wireless communication network includes, but is not limited to, 4G, 5G or LoRa, to ensure that the collected multi-source raw data can be synchronized to the ground control platform 100 in a timely and stable manner, supporting subsequent operations such as data preprocessing and feature extraction.

[0072] The first GPS positioning unit 230 is used to obtain the geographical coordinates of the water quality monitoring station 200, such as latitude and longitude, and associate and bind the positioning information with the raw data collected by the first data acquisition unit 210, so that the ground control platform 100 can clearly identify the specific monitoring point corresponding to each set of monitoring data, which is helpful in judging whether there are monitoring blind spots in the water quality monitoring area.

[0073] Among them, such as Figure 3 and Figure 4 As shown, the satellite remote sensing 300 is used to acquire information on water body distribution and water level changes in a large-scale water quality monitoring area, specifically including:

[0074] The second data acquisition unit 310 is used to collect electromagnetic wave reflection and radiation signals in the water quality monitoring area through optical sensors and radar sensors carried by the satellite. The signals include the spectral and morphological characteristics of the water body and surrounding materials. Based on the collected raw data, it extracts water body distribution information such as water body boundaries and water area, and combines different collected data to analyze the water level changes in the water quality monitoring area.

[0075] The second data transmission unit 320 is used to transmit the raw data acquired by the second data acquisition unit 310 to the ground control platform 100 through a satellite communication link, supporting the ground control platform's data preprocessing, feature extraction, and water quality monitoring model construction operations.

[0076] The second GPS positioning unit 330 is used to record the satellite orbit parameters, observation time and geographical coordinates of the corresponding ground monitoring area when remote sensing data is collected using the Global Positioning System. The positioning information is associated and bound with the raw data obtained by the second data acquisition unit 310, so that the ground control platform 100 can clearly identify the specific spatial location corresponding to each set of satellite remote sensing data, and ensure that the water distribution and water level change information are accurately matched to the monitoring area.

[0077] Among them, such as Figure 3 and Figure 4 As shown, the mobile monitoring equipment 400 includes unmanned aerial vehicle (UAV) equipment and unmanned surface vessel (USV) equipment, used for real-time on-site water quality measurement operations, specifically including:

[0078] The third data acquisition unit 410 includes a UAV equipped with a multispectral camera and a gas sensor, used to collect images of the water area and volatile organic compound concentration data, and to capture the surface conditions of the water body and information on surrounding air pollutants in key suspicious areas. The UAV is equipped with a water quality sensor, used to collect water quality indicators and watershed parameters such as pH value, dissolved oxygen, total phosphorus, and total nitrogen from multiple water samples, filling the monitoring gaps in the monitoring blind areas of the water quality monitoring station 200 and satellite remote sensing 300.

[0079] The third data transmission unit 420 is used to transmit the raw data collected by the third data acquisition unit 410 to the ground control platform 100 in real time through a wireless communication network.

[0080] The third GPS positioning unit 430 is used to acquire the real-time position coordinates of the UAV equipment during flight, as well as the specific point coordinates of the unmanned vessel equipment when it moves to collect water samples in the water quality monitoring area. The positioning information is then associated and bound with the corresponding collected data, so that the ground control platform 100 can clearly identify the spatial location corresponding to each set of mobile monitoring data and assist in judging the spatial correlation of the monitoring data.

[0081] Among them, such as Figure 3 and Figure 4 As shown, meteorological station 500 is used to acquire meteorological data related to hydrology and water quality, specifically including:

[0082] The fourth data acquisition unit 510 is used to acquire meteorological data related to hydrology and water quality. It is equipped with a rain gauge, temperature and humidity sensor, wind speed sensor, visibility meter, and sunshine meter to collect meteorological data related to hydrology and water quality in real time. The rain gauge collects precipitation and converts it into an electrical signal for recording. The temperature and humidity sensor records the ambient temperature. The wind speed sensor calculates the wind speed using the ultrasonic principle. The visibility meter measures the weather visibility.

[0083] The fourth data transmission unit 520 is used to transmit the meteorological data acquired by the fourth data acquisition unit 510 to the ground control platform 100 through a wireless communication network, so as to provide real-time data input for the weight adjustment module 130 to dynamically adjust the individual weights and multi-device collaborative weights of each monitoring device, as well as for the water quality monitoring model to analyze the impact of meteorology on water quality.

[0084] The fourth GPS positioning unit 530 is used to obtain the geographical coordinates of the meteorological station 500 and associate the positioning information with the meteorological data collected by the fourth data acquisition unit 510, so that the ground control platform 100 can clearly identify the specific spatial location corresponding to each set of meteorological data, and then accurately match the raw water quality data collected by the water quality monitoring station 200, mobile monitoring equipment 400, etc. in the area.

[0085] Example 2

[0086] Among them, such as Figure 2 As shown, unlike Example 1, this example describes the operation flow of a corresponding module in an intelligent water quality monitoring system based on multi-source data fusion. The process of feature extraction module 120 extracting key feature data specifically includes:

[0087] Monitoring data is extracted from the initial dataset, the correlation of variables in the initial dataset is analyzed, and key feature data with high correlation are identified. Key feature data include collection source information, time and location, water quality indicators, watershed parameters and meteorological data.

[0088] For feature extraction of the collected source information, the unique identifier of the data source is determined by recording the equipment number, type and monitoring network node information of each monitoring device, and the stability parameters of the data transmission link are also recorded.

[0089] For the extraction of time and location features, the latitude and longitude coordinates of each monitoring point are obtained through the GPS positioning unit built into the monitoring device. Combined with the electronic map data of the water quality monitoring area, the latitude and longitude coordinates are converted into watershed partition codes. The data collection time is recorded using timestamps, and the time dimension features including year, month, day, hour and minute are decomposed and marked as whether it is a peak period for water quality monitoring.

[0090] For the feature extraction of water quality indicators, pH value is obtained by measuring the hydrogen ion concentration in the water body using a pH sensor, dissolved oxygen content in the water body is detected by a dissolved oxygen sensor, spectral features related to algae are extracted from the spectral data of remote sensing data to reflect the density and species distribution of algae in the water body, images and video data are captured by a high-precision camera, and edge detection and texture analysis technology is used to identify algae species and density, and quantitative data of total bacterial count and specific bacterial groups are obtained through laboratory PCR amplification technology.

[0091] For the feature extraction of watershed parameters, the water level in the water body is measured by a water level gauge, the flow velocity in the water quality monitoring area is measured by a flow meter, the particle size distribution and concentration of suspended particulate matter in the water body are measured by a laser particle size analyzer, the turbidity value of the water body is obtained by a turbidity sensor, and pollution indicators including ammonia nitrogen concentration, total phosphorus and total nitrogen are detected by a water quality analyzer. Geographic feature parameters including watershed slope, vegetation coverage and water area are extracted by combining topographic mapping data.

[0092] For feature extraction of meteorological data, precipitation is collected by rain gauges and converted into electrical signals to record rainfall; temperature and humidity sensors are used to record the temperature and relative humidity of the surrounding environment; wind speed is calculated using wind speed sensors; sunshine duration and intensity are recorded by sunshine meters; and weather visibility is measured by visibility meters.

[0093] All the key feature data extracted above were standardized to eliminate dimensional differences. The correlation between features was analyzed using Pearson correlation coefficient to identify highly correlated feature combinations. PCA dimensionality reduction technology was applied to reduce the number of features and retain principal components. Principal components with a cumulative contribution rate of more than 90% were retained as input features for subsequent models.

[0094] Among them, such as Figure 2 As shown, the weight adjustment module 130 adopts a hierarchical weight calculation method. The hierarchical weight includes individual weights and multi-device collaborative weights. Individual weights are used for a single type of monitoring device. Based on the differences in the monitoring focus of its monitoring points under different weather conditions, environments, and watersheds, the weights of the monitoring parameters within the corresponding monitoring device are adjusted. For example, when the visibility is high on a clear day (i.e., when the macro-pollution distribution monitoring accuracy is high), the individual weight of its image parameter is increased. When the visibility is low on a rainy day (i.e., when the visibility is low), the individual weight of this parameter is decreased to avoid the expansion of data errors due to environmental interference and to ensure the effectiveness of the output data of a single monitoring device. In terms of specificity and targeting, the multi-device collaborative weighting is used to adjust the parameter weight ratio of each monitoring device in the model construction based on the differences in monitoring parameters of different monitoring devices when all monitoring devices are monitoring the same target object individually. For example, when monitoring a large open water quality monitoring area, the coverage of satellite remote sensing 300 is high, so its multi-device collaborative weighting is increased to dominate the judgment of macro water quality. When monitoring complex nearshore waters with many monitoring blind spots, the water sample collection accuracy of unmanned surface vessel equipment is high, so its collaborative weighting is increased to supplement micro-pollution index data. Ultimately, this achieves complementary advantages of multi-source data and improves the accuracy and reliability of the fused data.

[0095] Furthermore, the formula for calculating the individual weight is as follows:

[0096] ;

[0097] in, For the first The first type of monitoring device Individual weights of each monitoring parameter, , , and All are weighted parameters. It can be adapted and adjusted according to actual needs. For the first The first type of monitoring device The basic weights of each monitoring parameter This is a function of the meteorological influence coefficient. For environmental impact coefficient function, The location influence coefficient function allows for the acquisition of basic weights, monitoring coverage, and monitoring accuracy from equipment factory parameters, historical calibration data, and field measurements. Meteorological, environmental, and location influence coefficients can be preset within ranges using meteorological data, watershed pollution archives, and topographic maps of the monitoring area. No complex calculations are required. When the scenario changes, only local parameters such as meteorological and environmental influence coefficients need to be adjusted to quickly recalculate the weights without reconstructing the formula, thus adapting to real-time monitoring needs. (Meteorological influence coefficient function) Used to quantify meteorological conditions on the first The impact of the accuracy of each monitoring parameter is considered, based on a comparison of real-time meteorological data with preset thresholds, and the location influence coefficient function. Environmental impact coefficient function used to quantify the impact of specific geographical location characteristics of monitoring points. The function is used to quantify the impact of the surrounding environment of the water body on monitoring parameters. It can be preset based on geographic information system (GIS) data. The above function can be generated based on historical data and expert experience.

[0098] Example: In a certain water quality monitoring area, satellite remote sensing 300 is used to monitor algae density. This parameter is easily affected by meteorological conditions, environmental factors, and location. Assume the factory-calibrated basic weights for satellite remote sensing algae density monitoring. The weight is 0.7, and the weighting parameter is set according to the actual monitoring needs, assuming it is 0.7. , , and For optical imagery data from Satellite Remote Sensing 300, the meteorological impact coefficient function, with visibility as the core indicator, can be assumed to be: Environmental impact coefficient function Using water turbidity as an indicator, we can assume that: The location influence coefficient function uses the type of monitoring area as an indicator, and can be assumed to be: Assuming the scenario has a visibility of 12km, a water turbidity of 3 NTU, and is located in an open, main waterway, then... The value is higher than the basic weight, which reflects the improvement of data reliability brought by "sunny day + clear and open water". This parameter should be given priority when merging data in the future.

[0099] The formula for calculating the collaborative weight of multiple devices is:

[0100] ;

[0101] in, For the first Multi-device collaborative weighting of similar monitoring devices For the first The monitoring coverage of this type of monitoring device For the first The monitoring accuracy of this type of monitoring device The number of monitoring device types participating in collaborative monitoring. For all Monitoring devices that participate in collaborative monitoring Perform a summation operation. Monitoring coverage For the first The proportion of the effective monitoring range of a monitoring device to the total area of ​​the target water quality monitoring area within a specific time period can be expressed by the formula: ,in, For the first The effective monitoring area of ​​this type of monitoring device The total area of ​​the target water quality monitoring zone, and the monitoring accuracy. For the first The accuracy of the data provided by such monitoring devices is usually set based on the device's factory calibration parameters, historical calibration data, or comparison results with standard measurement methods, and is a normalized value between 0 and 1.

[0102] Among them, such as Figure 2 As shown, the construction process of the water quality monitoring model in model construction module 150 specifically includes:

[0103] Specifically, the corresponding monitoring devices collect corresponding data, and the data preprocessing module 110 performs standardization processing on the received multi-source raw data. First, a normalization operation is performed to unify the representation, units, and format of data from different data sources. For example, dissolved oxygen, pH value, and other indicators are normalized. Then, data cleaning is performed to remove duplicate and abnormal data, outputting a well-organized initial dataset. The normalization formula is: , This is the raw data for a certain type of water quality data. For normalized water quality data, To correspond to the minimum value in a certain type of water quality data, To correspond to the maximum value in a certain type of water quality data, key feature data are extracted based on monitoring data from the initial dataset and the fused dataset. A feature dataset containing the extracted key feature data is constructed, and the feature dataset is divided into a training set and a test set. The feature dataset includes associated data of collection source information, time and location, water quality indicators, watershed parameters, and meteorological data. Examples of collection source information are: "satellite remote sensing - water body distribution data" and "unmanned vessel - total phosphorus data". An example of time and location is: "2024-05-20 14:30". Water quality indicators include core indicators such as pH value, dissolved oxygen, algae density, and total bacterial count. Watershed parameters include extracted water level, flow velocity, flow rate, and pollution level. Meteorological data includes rainfall, weather visibility, temperature, and wind speed.

[0104] The weight adjustment module 130 dynamically adjusts data weights for single monitoring devices such as satellite remote sensing 300 and unmanned surface vessels, combining meteorological, environmental, and location data through preset formulas. Adjust the weights of various parameters within the corresponding devices. When multiple devices monitor the same area, the monitoring coverage and accuracy of the devices are determined using a formula. The overall weight ratio of each device in the model is adjusted. The data fusion module 140 performs fusion processing on multi-source data based on the adjusted individual weights and multi-device collaborative weights. First, the contribution of internal parameters of each device is optimized according to the individual weights. Then, the data of different devices are integrated by combining the multi-device collaborative weights. For example, large water areas are mainly based on macro data from satellite remote sensing 300, while nearshore blind areas are mainly based on micro data from unmanned surface vessel equipment. Data redundancy and conflicts are eliminated to generate a fusion dataset with a unified structure.

[0105] Based on the needs of water quality monitoring, and combined with machine learning algorithms suitable for multi-source data, a water quality monitoring model is constructed using the training set. During the model construction process, the individual weights and multi-device collaborative weights obtained by the weight adjustment module 130 are incorporated to enhance the reasonable contribution of different data sources to the water quality monitoring results. The water quality monitoring model can be implemented based on existing technologies, such as random forest and PCA algorithms.

[0106] The monitoring accuracy of the water quality monitoring model was evaluated using a test set, and different performance indicators were calculated and compared.

[0107] The differences between the monitoring results of the water quality monitoring model and the actual water quality are analyzed to identify potential sources of error. The model parameters are then adjusted, and the trained water quality monitoring model is deployed to the production environment of the ground control platform 100 to receive updated multi-source fusion data in real time. The operation status and prediction results of the water quality monitoring model are displayed in a visual form, which intuitively shows the water quality health status of each monitoring area. Staff can identify water quality anomalies in real time through the output results, and at the same time, the model parameters are calibrated in reverse based on the actual water quality.

[0108] Furthermore, water quality indicators include pH value, dissolved oxygen, algae density and total bacterial count; watershed parameters include water level, flow velocity, suspended particulate matter concentration, turbidity, pollution level and geographical features; and meteorological data include rainfall, temperature and humidity, wind speed, solar radiation intensity and weather visibility.

[0109] In summary, this invention establishes an intelligent water quality monitoring system based on multi-source data fusion. By integrating multiple data acquisition methods such as water quality monitoring stations, satellite remote sensing, mobile monitoring equipment, and meteorological stations, it can capture water quality information in all aspects, improve the ability to capture water quality changes in complex watersheds, and effectively compensate for the shortcomings of single data sources. This enables all-time, multi-dimensional monitoring of water quality, overcomes the limitations of traditional single monitoring methods, and has good application prospects.

[0110] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An intelligent water quality monitoring system based on multi-source data fusion, characterized in that, The system includes a ground control platform (100) and monitoring devices that communicate with each other via a network. The ground control platform (100) is used to periodically collect and identify raw data from different sources and different water quality monitoring areas, and to process, analyze, and issue early warnings for the raw data. The monitoring devices are used to acquire multi-source raw data related to the water quality monitoring areas and to conduct comprehensive monitoring of scene information and environmental data of the water quality monitoring areas. The monitoring devices include a water quality monitoring station (200), a satellite remote sensing device (300), a mobile monitoring device (400), and a weather station (500). The water quality monitoring station (200), the satellite remote sensing device (300), the mobile monitoring device (400), and the weather station (500) all transmit the collected raw data to the ground control platform (100) via a wireless communication network. The ground control platform (100) includes: The data preprocessing module (110) is used to preprocess the collected multi-source raw data, perform normalization operations and remove duplicate and abnormal data, add timestamps, so that the water quality data from different data sources are consistent in terms of presentation, unit and format, and construct the initial dataset. The feature extraction module (120) is used to analyze the monitoring data in the initial dataset and extract key feature data; The weight adjustment module (130) is used to adjust the weight of each data source in real time based on meteorological data, parameters of the movement process of the monitoring device, and water quality indicators. The data fusion module (140) is used to fuse multi-source water quality data and integrate data from different sources and of different types to generate a fused dataset. The model building module (150) is used to build a water quality monitoring model based on the fused dataset and key feature data, and to monitor and output the operation status and prediction results of the water quality monitoring model in real time. The monitoring and evaluation module (160) is used to output the operation status and prediction results of the water quality monitoring model in a visual form, so as to facilitate the timely detection of water quality anomalies.

2. The intelligent water quality monitoring system based on multi-source data fusion according to claim 1, characterized in that, The water quality monitoring station (200) is equipped with various types of sensors to collect real-time water quality scene information and environmental data for the corresponding water quality monitoring area, specifically including: The first data acquisition unit (210) is equipped with a pH sensor, dissolved oxygen sensor, turbidity sensor, COD sensor, ammonia nitrogen sensor, detector and high-precision camera, used to collect water quality indicators and watershed parameters of the corresponding water quality monitoring area in real time; The first data transmission unit (220) is responsible for summarizing the raw data collected by several sensors and transmitting it to the ground control platform (100) through a wireless communication network. The first GPS positioning unit (230) is used to obtain the geographical coordinates of the water quality monitoring station (200) and associate and bind the positioning information with the raw data collected by the first data acquisition unit (210).

3. The intelligent water quality monitoring system based on multi-source data fusion according to claim 1, characterized in that, The satellite remote sensing (300) is used to acquire information on water body distribution and water level changes in a large-scale water quality monitoring area, specifically including: The second data acquisition unit (310) is used to acquire electromagnetic wave reflection and radiation signals of the water quality monitoring area through optical sensors and radar sensor equipment carried by the satellite. The signals include the spectral and morphological characteristics of the water body and the surrounding materials. The second data transmission unit (320) is used to transmit the raw data acquired by the second data acquisition unit (310) to the ground control platform (100) via a satellite communication link. The second GPS positioning unit (330) is used to record the satellite orbit parameters, observation time and geographical coordinates of the corresponding ground monitoring area during remote sensing data acquisition using the global positioning system, and to associate and bind the positioning information with the raw data acquired by the second data acquisition unit (310).

4. The intelligent water quality monitoring system based on multi-source data fusion according to claim 1, characterized in that, The mobile monitoring equipment (400) includes unmanned aerial vehicle (UAV) equipment and unmanned vessel equipment, used for real-time on-site water quality measurement operations, specifically including: The third data acquisition unit (410) is equipped with a multispectral camera and a gas sensor to collect images of the water area and volatile organic compound concentration data. The unmanned boat is equipped with a water quality sensor to collect water quality indicators and watershed parameters from multiple water samples. The third data transmission unit (420) is used to transmit the raw data collected by the third data acquisition unit (410) to the ground control platform (100) in real time through the wireless communication network. The third GPS positioning unit (430) is used to obtain the real-time position coordinates of the UAV equipment during flight, as well as the specific point coordinates of the unmanned vessel equipment when it moves to collect water samples in the water quality monitoring area, and to associate and bind the positioning information with the corresponding collected data.

5. The intelligent water quality monitoring system based on multi-source data fusion according to claim 1, characterized in that, The meteorological station (500) is used to acquire meteorological data related to hydrology and water quality, specifically including: The fourth data acquisition unit (510) is used to acquire meteorological data related to hydrology and water quality. It is equipped with rain gauges, temperature and humidity sensors, wind speed sensors, visibility detectors, and sunshine meters to collect meteorological data related to hydrology and water quality in real time. The fourth data transmission unit (520) is used to transmit the meteorological data acquired by the fourth data acquisition unit (510) to the ground control platform (100) through a wireless communication network. The fourth GPS positioning unit (530) is used to obtain the geographical coordinates of the meteorological station (500) and associate and bind the positioning information with the meteorological data collected by the fourth data acquisition unit (510).

6. The intelligent water quality monitoring system based on multi-source data fusion according to claim 1, characterized in that, The process of the feature extraction module (120) extracting key feature data specifically includes: Monitoring data is extracted from the initial dataset, the correlation of variables in the initial dataset is analyzed, and key feature data with high correlation are identified. The key feature data includes collection source information, time and location, water quality indicators, watershed parameters and meteorological data. For feature extraction of the collected source information, the unique identifier of the data source is determined by recording the equipment number, type and monitoring network node information of each monitoring device, and the stability parameters of the data transmission link are also recorded. For the extraction of time and location features, the latitude and longitude coordinates of each monitoring point are obtained through the GPS positioning unit built into the monitoring device. Combined with the electronic map data of the water quality monitoring area, the latitude and longitude coordinates are converted into watershed partition codes. The data collection time is recorded using timestamps, and the time dimension features including year, month, day, hour and minute are decomposed and marked as whether it is a peak period for water quality monitoring. For the feature extraction of water quality indicators, pH value is obtained by measuring the hydrogen ion concentration in the water body using a pH sensor, dissolved oxygen content in the water body is detected by a dissolved oxygen sensor, spectral features related to algae are extracted from the spectral data of remote sensing data to reflect the density and species distribution of algae in the water body, and algae species and density are identified by capturing images and video data with a high-precision camera and using edge detection and texture analysis technology. For the feature extraction of watershed parameters, the water level in the water body is measured by a water level gauge, the flow velocity in the water quality monitoring area is measured by a flow meter, the particle size distribution and concentration of suspended particulate matter in the water body are measured by a laser particle size analyzer, the turbidity value of the water body is obtained by a turbidity sensor, and pollution indicators including ammonia nitrogen concentration, total phosphorus and total nitrogen are detected by a water quality analyzer. Geographic feature parameters including watershed slope, vegetation coverage and water area are extracted by combining topographic mapping data. For feature extraction of meteorological data, precipitation is collected by rain gauges and converted into electrical signals to record rainfall; temperature and humidity sensors are used to record the temperature and relative humidity of the surrounding environment; wind speed is calculated using wind speed sensors; sunshine duration and intensity are recorded by sunshine meters; and weather visibility is measured by visibility meters. All the key feature data extracted above were standardized, and the correlation between features was analyzed using Pearson correlation coefficient to identify highly correlated feature combinations. PCA dimensionality reduction technique was applied to reduce the number of features and retain principal components.

7. The intelligent water quality monitoring system based on multi-source data fusion according to claim 1, characterized in that, The weight adjustment module (130) adopts a hierarchical weight calculation method. The hierarchical weight includes individual weight and multi-device collaborative weight. The individual weight is used to adjust the weight of each monitoring parameter within a single type of monitoring device according to the difference in the monitoring focus of its monitoring point under different weather, environment and watershed conditions. The multi-device collaborative weight is used to adjust the parameter weight ratio of each monitoring device in the model construction according to the difference in the monitoring parameters of different monitoring devices when all monitoring devices are monitoring the same target object.

8. The intelligent water quality monitoring system based on multi-source data fusion according to claim 7, characterized in that, The formula for calculating the individual weight is as follows: ； in, For the first The first type of monitoring device Individual weights of each monitoring parameter, , , and All are weighted parameters. It can be adapted and adjusted according to actual needs. For the first The first type of monitoring device The basic weights of each monitoring parameter This is a function of the meteorological influence coefficient. For environmental impact coefficient function, This is a function of the location influence coefficient; The formula for calculating the multi-device collaborative weight is as follows: ； in, For the first Multi-device collaborative weighting of similar monitoring devices For the first The monitoring coverage of this type of monitoring device For the first The monitoring accuracy of this type of monitoring device The number of monitoring device types participating in collaborative monitoring. For all Monitoring devices that participate in collaborative monitoring Perform a summation operation. .

9. The intelligent water quality monitoring system based on multi-source data fusion according to claim 6, characterized in that, The construction process of the water quality monitoring model in the model construction module (150) specifically includes: Based on the monitoring data in the initial dataset and the fused dataset, key feature data are extracted, and a feature dataset containing the extracted key feature data is constructed. The feature dataset is divided into a training set and a test set. The feature dataset includes the associated data of the collection source information, time and location, water quality indicators, watershed parameters and meteorological data. Based on the water quality monitoring requirements, combined with machine learning algorithms suitable for multi-source data, a water quality monitoring model is constructed using the training set. In the process of model construction, the individual weights and multi-device collaborative weights obtained by the weight adjustment module (130) are incorporated to strengthen the reasonable contribution of different data sources to the water quality monitoring results. The monitoring accuracy of the water quality monitoring model was evaluated using a test set, and different performance indicators were calculated and compared. The differences between the monitoring results of the water quality monitoring model and the actual water quality are analyzed, potential sources of error are identified, and the model parameters are adjusted. The trained water quality monitoring model is then deployed to the production environment of the ground control platform (100) to receive updated multi-source fusion data in real time, as well as the operation status and prediction results of the water quality monitoring model.

10. The intelligent water quality monitoring system based on multi-source data fusion according to claim 9, characterized in that, The water quality indicators include pH value, dissolved oxygen, algae density and total bacterial count. The watershed parameters include water level, flow velocity, suspended particulate matter concentration, turbidity, pollution level and geographical features. The meteorological data include rainfall, temperature and humidity, wind speed, solar radiation intensity and weather visibility.

Images