Direct drinking water microorganism real-time monitoring and early warning system based on multi-source data fusion

Abstract

The invention provides a direct drinking water microorganism real-time monitoring and early warning system based on multi-source data fusion, and relates to the technical field of direct drinking water quality monitoring, and the system comprises the following modules: a data acquisition module, a multi-source sensor network installed in a direct drinking water pipeline according to the monitoring requirement of direct drinking water, a data transmission module, and a real-time monitoring and early warning module employing a LoRa / wiFi / 5G hybrid communication protocol, an appropriate communication mode is selected according to different scenes and requirements, and data acquired by the data acquisition module is transmitted to a cloud or an edge computing node through an MQTT protocol; the data processing and analysis module is used for carrying out preprocessing, multi-source data fusion and dynamic prediction and early warning on the data transmission module, and the user interaction and decision module is provided with a Web / APP visual interface; multi-source data such as microbial fluorescence spectrum, water quality parameters, pipe network topology and meteorology are integrated, microbial pollution of direct drinking water is monitored from multiple dimensions, and monitoring comprehensiveness and accuracy are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of direct drinking water quality monitoring, and particularly relates to a direct drinking water microorganism real-time monitoring and early warning system based on multi-source data fusion. BACKGROUND

[0002] With the acceleration of urbanization and the improvement of public drinking water safety awareness, direct drinking water systems have become an important part of modern urban infrastructure; however, water pollution, especially microbial pollution, is becoming increasingly prominent, such as the presence of E. coli, Pseudomonas aeruginosa, etc. in water bodies not only affects water quality, but also poses a serious threat to public health.

[0003] Traditional water quality monitoring methods usually rely on offline detection in the laboratory, which has the defects of low timeliness, inability to realize real-time early warning, etc., and is difficult to meet the efficient monitoring needs of modern cities for water quality safety; and the traditional detection method often cannot capture early pollution signals in time. SUMMARY

[0004] The purpose of the present application is to provide a direct drinking water microorganism real-time monitoring and early warning system based on multi-source data fusion, which solves the technical problems existing in the prior art.

[0005] In order to achieve the above-mentioned purpose of the application, the technical scheme adopted by the present application is as follows:

[0006] A direct drinking water microorganism real-time monitoring and early warning system based on multi-source data fusion comprises the following modules: a data acquisition module, which installs a multi-source sensor network in the direct drinking water pipeline according to the monitoring needs of the direct drinking water, collects multi-source data of water quality parameters, microorganism data and environmental parameters of the direct drinking water; collects and organizes pipe network topology structure data, accurately obtains the layout, pipe diameter and node position information of the pipe network through field survey and drawing, and then establishes a historical pollution event database to record detailed information of past microbial pollution events; a data transmission module, which adopts a LoRa / wiFi / 5G hybrid communication protocol, selects a suitable communication mode according to different scenes and needs, and transmits the data collected by the data acquisition module to the cloud or an edge computing node through an MQTT protocol; a data processing and analysis module, which pre-processes, multi-source data fuses and dynamically predicts and warns the data transmission module; the multi-source data fusion comprises: feature-level fusion: an improved canonical correlation analysis algorithm is used to project the microorganism fluorescence signal and the water quality parameter into a shared space to generate a joint confidence matrix to solve the sensor conflict problem; model-level fusion: a spatio-temporal graph convolution network is constructed to fuse time series data and spatial data and capture the microorganism diffusion law; a user interaction and decision module, which is provided with a Web / APP visualization interface and supports real-time data display, pollution heat map tracking and emergency decision suggestions.

[0007] Further, the multi-source sensor network in the data acquisition module includes: a microorganism detection unit deploying an ATP biological fluorescence sensor for detecting microorganism metabolic activity signals, and combining ultraviolet-visible spectrum and three-dimensional fluorescence spectrum sensors for enhancing the identification ability of microorganisms; a water quality parameter unit including a pH sensor, a dissolved oxygen sensor, a residual chlorine sensor and a turbidity sensor for real-time and accurate monitoring of the chemical index changes of direct drinking water; an environmental perception unit collecting temperature, flow rate and pipe network pressure data, while integrating temperature and humidity, rainfall data provided by a weather station and regional water quality thermal map data obtained by satellite remote sensing to construct a space-time correlation model.

[0008] Further, the preprocessing in the data processing and analysis module includes: using a box plot method to remove outliers from the data collected by the data acquisition module, and removing unreasonable data caused by sensor failure or external interference and other factors; using a Kalman filter algorithm to perform noise reduction processing on the data; and simultaneously, using a normalization method to unify the data scale.

[0009] Further, the dynamic prediction and early warning in the data processing and analysis module includes: based on an LSTM-Transformer hybrid model, using a long short-term memory network to capture the time sequence dependence of microorganism concentration change data and learn long-term dependence relationships in the data; simultaneously, combining a Transformer architecture to extract global features of multi-sensor data and focus on the mutual relationship between different sensor data; combining the above two, and then predicting the microorganism risk index in the next 12 hours; self-adaptive threshold adjustment, dynamically optimizing the early warning threshold according to environmental parameters and historical data.

[0010] Further, the user interaction and decision module includes: an intelligent tracing unit calculating the probability distribution of the pollution source, quickly locating the high-risk node and determining the possible pollution source position by combining the pipe network topology structure through a Bayesian network and according to the monitoring data of each node when the pollution event occurs; an emergency response unit automatically triggering ultraviolet disinfection equipment to disinfect direct drinking water when detecting microorganism pollution abnormalities, and simultaneously closing the pollution branch valve; and then generating a detailed disposal report and pushing it to the management personnel.

[0011] The present application integrates multi-source data such as microbial fluorescence spectrum, water quality parameters, pipe network topology and weather, monitors the microbial pollution of direct drinking water from multiple dimensions, improves the comprehensiveness and accuracy of monitoring, and can more timely and accurately find potential microbial pollution problems; secondly, a micro AI chip is deployed at the sensor end to realize data preprocessing and preliminary modeling, reduce data transmission volume, reduce cloud load, improve system response speed and real-time performance, and can quickly process and analyze part of the data locally to discover abnormal conditions in time; and real-time threshold optimization is combined with environmental parameters to avoid the limitations of traditional fixed thresholds; in addition, through the fusion of spatio-temporal graph convolution and Bayesian network, the pollution source is quickly located, and the response time is shortened to within 10 minutes. BRIEF DESCRIPTION OF DRAWINGS

[0012] Figure 1 The flowchart of the present application. DETAILED DESCRIPTION

[0013] In order to make the content of the present application easier to be clearly understood, the technical solutions in the embodiments of the present application will be clearly and completely described below in combination with the accompanying drawings in the embodiments of the present application. The same parts are denoted by the same reference numerals. It should be noted that the words "front", "back", "left", "right", "up" and "down" used in the following description refer to the directions in the drawings, and the words "inner" and "outer" refer to the directions towards or away from the geometric center of a particular part.

[0014] As shown in Figure 1 The present application provides a direct drinking water microbial real-time monitoring and early warning system based on multi-source data fusion, which includes the following modules:

[0015] A data acquisition module is installed with a multi-source sensor network in a direct drinking water pipeline according to the monitoring requirements of direct drinking water, and multi-source data of water quality parameters, microbial data and environmental parameters of direct drinking water are collected; a micro AI chip such as Huawei Ascend is deployed at the end of the multi-source sensor to realize data preprocessing and preliminary modeling, and reduce the cloud load; this module collects and organizes pipe network topology data, accurately obtains the layout, pipe diameter and node position information of the pipe network through field surveying and drawing, and then establishes a historical pollution event database to record detailed information of past microbial pollution events, and provides a reference basis for modeling of pollution diffusion path, so as to quickly analyze the possible transmission path when a pollution event occurs; the multi-source sensor network in the data acquisition module includes: a microbial detection unit, which deploys an ATP biological fluorescence sensor for detecting microbial metabolic activity signals, and combines ultraviolet-visible spectrum and three-dimensional fluorescence spectrum sensors for enhancing the recognition ability of microorganisms and accurately capturing characteristic information of microorganisms such as humic acid fluorescence peak and tryptophan fluorescence peak as microbial activity indicators; a water quality parameter unit, which includes pH sensors, dissolved oxygen sensors, residual chlorine sensors and turbidity sensors for real-time and accurate monitoring of chemical index changes of direct drinking water, and the above chemical index changes are closely related to the growth and reproduction of microorganisms, and through monitoring of the above indexes, multi-dimensional data support can be provided for analysis of microbial pollution; an environmental perception unit acquires temperature, flow rate and pipe network pressure data, integrates temperature and humidity, rainfall data provided by a weather station and regional water quality thermal map data obtained by satellite remote sensing, constructs a space-time correlation model, comprehensively considers the influence of time and space factors on the growth and transmission of microorganisms, and comprehensively masters the environmental conditions of the direct drinking water system;

[0016] A data transmission module adopts a LoRa / wiFi / 5G hybrid communication protocol, selects a suitable communication mode according to different scenes and requirements, transmits data collected by the data acquisition module to the cloud or edge computing nodes through the MQTT protocol, ensures timely and accurate transmission of data, and provides protection for subsequent data processing and analysis;

[0017] The data processing and analysis module performs preprocessing, multi-source data fusion, and dynamic prediction and early warning on the data transmission module. The preprocessing includes: using box plots to remove outliers from the data acquired by the data acquisition module, eliminating unreasonable data caused by sensor malfunctions or external interference; specifically, background subtraction, scattering correction, and smoothing are performed on the microbial fluorescence spectra, and characteristic peaks are extracted as indicators of microbial activity; water quality parameters are extracted using the sliding window method to extract time-series characteristics; and the Kalman filter algorithm is used to reduce noise in the data, improving data quality and stability. Simultaneously, a normalization method is used to unify the data scale, making data from different types of sensors more comparable, thus facilitating subsequent multi-source data fusion and dynamic prediction and early warning. Laying the foundation; the aforementioned multi-source data fusion includes: Feature-level fusion: using an improved canonical correlation analysis algorithm, i.e., CCA, microbial fluorescence signals and water quality parameters are projected into a shared space to generate a joint confidence matrix. By calculating the projection matrix of multi-sensor data, joint confidence moments are generated to resolve sensor conflict issues, uncover potential correlations between different data sources, and extract more representative fusion feature vectors. Specifically, the improved canonical correlation analysis algorithm can perform dimensionality reduction processing on the collected raw data, using principal component analysis or independent component analysis. If principal component analysis, i.e., PCA, is used, it projects the raw data into a low-dimensional space by performing a linear transformation on the data, while retaining the main features of the data. PCA is then performed on the two sets of variables separately. After dimensionality reduction, CCA analysis is performed to calculate the covariance matrix and perform eigenvalue decomposition in the low-dimensional space, which greatly reduces the amount of computation and improves the efficiency of the algorithm. Model-level fusion: a spatiotemporal graph convolutional network is constructed based on a deep learning framework to transform the pipe network topology into a graph structure. With pipe network nodes as vertices, edge weights are determined according to the direction and distance of water flow, and then time-series data and spatial data are fused to capture the microbial diffusion pattern. The dynamic prediction and early warning include: based on the LSTM-Transformer hybrid model, the preprocessed and fused data is input into the model, and the long short-term memory network is used to capture the temporal dependence of microbial concentration change data, learn the long-term dependence in the data, and capture the trend of microbial concentration change over time. Simultaneously, the system extracts global features from multi-sensor data using the Transformer architecture, focusing on the interrelationships between different sensor data. Combining these two approaches, it predicts the microbial risk index for the next 12 hours, providing quantitative indicators for early warning. Adaptive threshold adjustment is implemented by establishing a mathematical model based on environmental parameters and historical data. Analysis of historical monitoring data determines historical baseline values, and the impact of environmental parameters on microbial growth is used to determine parameters such as temperature influencing factors. During actual operation, environmental parameter data is acquired in real time to adjust and dynamically optimize the early warning threshold. Furthermore, an evaluation mechanism for threshold adjustment is established, continuously optimizing the model by comparing actual pollution conditions with early warning results, thereby improving the accuracy and reliability of early warnings.

[0018] The user interaction and decision-making module features a web / app visual interface, supporting real-time data display, pollution heat map tracking, and emergency decision-making suggestions. The pollution heat map tracking function displays the microbial contamination status of different areas in the drinking water system via a map, allowing users to intuitively understand the contamination distribution. The emergency decision-making suggestions provide users with targeted emergency measures based on early warning information and historical treatment experience, including disinfection equipment activation methods and valve closure strategies. The user interaction and decision-making module includes: an intelligent source tracing unit, which uses a Bayesian network combined with the pipeline topology to calculate the probability distribution of pollution sources based on monitoring data from each node at the time of a pollution event, quickly locating high-risk nodes and determining possible pollution source locations; and an emergency response unit, which automatically triggers ultraviolet disinfection equipment to disinfect the drinking water when abnormal microbial contamination is detected, while simultaneously closing the valves of the contaminated branch to prevent contamination spread; then, a detailed handling report is generated and pushed to management personnel.

[0019] Example 1:

[0020] Specifically, this system will be installed in the urban drinking water supply area. First, a multi-source sensor network will be installed in the drinking water pipeline to ensure comprehensive and accurate collection of multi-source information such as water quality parameters, microbial data, and environmental parameters of the drinking water. At the same time, the results of this system will be compared with laboratory tests, and drinking water samples will be collected regularly and sent to a professional laboratory for microbial testing as a reference standard to verify the monitoring results of this system.

[0021] The accuracy of the system's prediction of microbial contamination was evaluated by comparing the results with those from laboratory tests. Over a period of 6 months, the system's predictions were consistent with the laboratory test data, and the system's prediction accuracy reached 93.5%, with the false alarm rate reduced to below 5%.

[0022] In summary, this system can accurately predict the microbial contamination of drinking water, providing a reliable basis for timely preventive measures.

[0023] Example 2:

[0024] This system can also be widely used in community drinking water stations, hospitals, schools and other scenarios. By comparing the prediction results with laboratory test data, the prediction accuracy of this system reached 94% within 6 months, and the false alarm rate was reduced to below 4%.

[0025] This invention promotes a shift in water quality supervision from passive response to proactive prevention and control. In the future, this invention can be combined with blockchain technology to achieve data immutability, further enhance the credibility of supervision, and provide core support for water security in smart cities.

[0026] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A real-time monitoring and early warning system for microorganisms in drinking water based on multi-source data fusion, characterized in that: Includes the following modules: The data acquisition module installs a multi-source sensor network in the drinking water pipeline according to the monitoring needs of drinking water, and collects multi-source data on water quality parameters, microbial data and environmental parameters of drinking water. Collect and organize pipeline network topology data, accurately obtain pipeline layout, pipe diameter, and node location information through on-site surveys and drawing, and then establish a historical pollution event database to record detailed information on past microbial pollution events; The data transmission module adopts a hybrid communication protocol of LoRa / wiFi / 5G, selects the appropriate communication method according to different scenarios and needs, and transmits the data collected by the data acquisition module to the cloud or edge computing node through the MQTT protocol. The data processing and analysis module performs preprocessing, multi-source data fusion, and dynamic prediction and early warning on the data transmission module. The above-mentioned multi-source data fusion includes: Feature-level fusion: An improved canonical correlation analysis algorithm is used to project microbial fluorescence signals and water quality parameters into a shared space to generate a joint confidence matrix, thereby resolving the sensor conflict problem; Model-level fusion: Constructing a spatiotemporal graph convolutional network to fuse temporal and spatial data and capture the patterns of microbial diffusion; The user interaction and decision-making module features a web / app visual interface that supports real-time data display, pollution heat map tracking, and emergency decision-making suggestions.

2. The real-time monitoring and early warning system for microorganisms in direct drinking water based on multi-source data fusion according to claim 1, characterized in that: The multi-source sensor network in the data acquisition module includes: The microbial detection unit is equipped with an ATP biofluorescence sensor, which is used to detect microbial metabolic activity signals. It also combines a UV-Vis spectroscopy and a three-dimensional fluorescence spectroscopy sensor to enhance the identification ability of microorganisms. The water quality parameter unit, including pH sensor, dissolved oxygen sensor, residual chlorine sensor and turbidity sensor, monitors changes in chemical indicators of drinking water in real time and accurately. The environmental sensing unit collects temperature, flow rate, and pipeline pressure data, and integrates temperature, humidity, and rainfall data provided by meteorological stations with regional water quality heat map data obtained by satellite remote sensing to construct a spatiotemporal correlation model.

3. The real-time monitoring and early warning system for microorganisms in direct drinking water based on multi-source data fusion according to claim 1, characterized in that: The preprocessing in the data processing and analysis module includes: using box plots to remove outliers from the data acquired by the data acquisition module, eliminating unreasonable data caused by sensor malfunctions or external interference; using Kalman filtering to reduce noise in the data; and simultaneously, using normalization to standardize the data scale.

4. The real-time monitoring and early warning system for microorganisms in direct drinking water based on multi-source data fusion according to claim 3, characterized in that: The dynamic prediction and early warning in the data processing and analysis module includes: Based on the LSTM-Transformer hybrid model, the Long Short-Term Memory network is used to capture the temporal dependence of microbial concentration change data and learn the long-term dependencies in the data. At the same time, the global features of multi-sensor data are extracted by combining the Transformer architecture to focus on the interrelationships between different sensor data. The two are combined to predict the microbial risk index for the next 12 hours. Adaptive threshold adjustment: The warning threshold is dynamically optimized based on environmental parameters and historical data.

5. The real-time monitoring and early warning system for microorganisms in direct drinking water based on multi-source data fusion according to claim 1, characterized in that: The user interaction and decision-making module includes: The intelligent source tracing unit, through Bayesian network combined with pipeline topology, calculates the probability distribution of pollution sources based on monitoring data of each node when a pollution event occurs, quickly locates high-risk nodes, and determines the possible location of pollution sources. When an abnormal microbial contamination is detected, the emergency response unit automatically triggers the ultraviolet disinfection equipment to disinfect the drinking water and simultaneously shuts off the valve of the contaminated branch. A detailed handling report is then generated and sent to the management personnel.

Images