Real-time monitoring method and system for water environment by fusing multi-source sensing data and storage medium

Abstract

This application relates to the field of water quality monitoring and pollution early warning technology, and in particular to a method, system, and storage medium for real-time water environment monitoring that integrates multi-source sensor data. The method includes: generating a standardized water quality parameter sequence using historical data from multi-source sensors; determining the pollutant migration time window and extracting and decomposing the corresponding time series to obtain short-term fluctuation and long-term trend components; constructing a time-delay causal correlation network based on the components and analyzing the causal connection strength between parameters; identifying potential early signs through the causal network and generating preliminary anomaly alarms; weighting and correcting the short-term fluctuation components based on the alarms to identify the impact effects; combining the corrected components with the long-term trend to generate a risk baseline, and then integrating the impact effects to obtain a comprehensive pollution risk distribution sequence, thereby generating a monitoring guidance report. This application effectively improves the accuracy and timeliness of real-time water environment monitoring and early warning.

Description

Technical Field

[0001] This application relates to the field of water quality monitoring and pollution early warning technology, and in particular to a method, system and storage medium for real-time water environment monitoring that integrates multi-source sensor data. Background Technology

[0002] Water environment monitoring is a crucial foundation for ecological environmental protection and water resource management, playing a vital role in ensuring drinking water safety, maintaining the balance of aquatic ecosystems, and preventing sudden water pollution incidents. With the rapid development of sensor technology and the Internet of Things, real-time monitoring methods based on multi-source sensor data have gradually become a research hotspot in the field of water environment monitoring. By deploying multi-parameter sensors upstream and downstream of rivers, key water quality indicators such as pH, conductivity, and dissolved oxygen can be collected in real time, providing data support for water quality assessment and pollution early warning.

[0003] However, existing water environment monitoring methods still have significant shortcomings when dealing with complex river environments. On the one hand, existing methods typically analyze data from each monitoring point independently, failing to fully consider the impact of differences in upstream and downstream hydrological conditions on pollutant migration processes, such as flow velocity changes caused by water level differences and pollutant retention due to river meandering. This leads to inaccurate pollution source location and difficulty in tracing migration paths. On the other hand, existing methods lack in-depth exploration of the time-delay causal relationships between multiple parameters, making it difficult to detect the correlation between rapid decreases in pH and abnormal fluctuations in conductivity in the early stages of pollution, often missing the optimal early warning opportunity.

[0004] A deeper technical challenge lies in the fact that pollutants are affected by factors such as sediment release, water flow disturbance, and water reoxygenation during their migration. Pollution signals may be amplified or attenuated during spatial transmission, and existing methods have failed to effectively correct for this signal distortion, leading to discrepancies between risk assessment results at downstream monitoring points and the actual situation. Therefore, how to integrate multi-source sensor data, fully considering upstream and downstream hydrological differences, pollutant migration time lags, and signal transmission distortion, to achieve accurate real-time monitoring of the aquatic environment and effective assessment of pollution risks is a pressing technical problem that needs to be solved. Summary of the Invention

[0005] This application provides a method, system, and storage medium for real-time water environment monitoring that integrates multi-source sensor data, to improve the accuracy and timeliness of real-time water environment monitoring and early warning.

[0006] In a first aspect, this application provides a method for real-time monitoring of the water environment by fusing multi-source sensor data, the method comprising: S1. Use multi-source sensors to acquire historical records of upstream and downstream monitoring points, process the historical records, and generate a standardized water quality parameter sequence. S2. Based on the standardized water quality parameter sequence, determine the migration time window of pollutants from upstream to downstream; S3. Based on the migration time window, extract the time series, process the time series using the seasonal trend decomposition method, and obtain the short-term fluctuation component and the long-term trend component. S4. Based on the short-term fluctuation component and the long-term trend component, construct a time-delay causal relationship network to determine the causal connection strength between each water quality parameter; S5. Based on the time-delay causal relationship network, determine whether there are potential signals of early signs of pollution and generate preliminary abnormal alarm records; S6. Based on the preliminary abnormal alarm record, perform weighted correction on the short-term fluctuation component, and identify the impact effect based on the weighted correction result; S7. Based on the weighted and corrected short-term fluctuation component and the long-term trend component, a risk baseline sequence is generated. Further, combined with the impact effect, a comprehensive pollution risk distribution sequence is generated. Based on the comprehensive pollution risk distribution sequence, a monitoring work guidance report is formed.

[0007] Secondly, this application provides a real-time water environment monitoring system that integrates multi-source sensor data, the system comprising: The data processing module is used to acquire historical records of upstream and downstream monitoring points using multi-source sensors, process the historical records, and generate standardized water quality parameter sequences. The window definition module is used to determine the migration time window of pollutants from upstream to downstream based on a standardized water quality parameter sequence; The time series decomposition module is used to extract time series based on migration time windows, process the time series through seasonal trend decomposition methods, and obtain short-term fluctuation components and long-term trend components. The network construction module is used to construct a time-delay causal relationship network based on short-term fluctuation components and long-term trend components, and to determine the causal connection strength between various water quality parameters. The signal recognition module is used to determine whether there are potential signals of early signs of pollution based on time-delay causal correlation networks, and generate preliminary abnormal alarm records. The weighted correction module is used to perform weighted correction on the short-term fluctuation components based on the preliminary abnormal alarm records, and to identify the impact effect based on the weighted correction results. The risk assessment module is used to generate a risk baseline sequence based on the weighted and corrected short-term fluctuation components and long-term trend components. It further combines the impact effects to generate a comprehensive pollution risk distribution sequence, and forms a monitoring work guidance report based on the comprehensive pollution risk distribution sequence.

[0008] Thirdly, this application provides a computer device, including: a memory and a processor, wherein the memory stores machine-readable instructions executable by the processor, and when the computer device is running, the processor communicates with the memory via a bus, and when the machine-readable instructions are executed by the processor, the steps of the above-described method for real-time monitoring of the water environment by fusing multi-source sensor data are performed.

[0009] Fourthly, this application provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the aforementioned method for real-time monitoring of the water environment by fusing multi-source sensor data.

[0010] Compared with the prior art, the beneficial effects of the present invention are at least as follows: I. By introducing the influence mechanism of flow velocity difference caused by upstream and downstream water level difference and river section curvature on pollutant residence time, the historical records are processed in a targeted manner, effectively eliminating the interference of physical environmental factors on monitoring data, improving the accuracy and consistency of water quality parameter series, and providing a reliable data foundation for subsequent analysis.

[0011] Second, by combining the time series cross-correlation function with the seasonal trend decomposition method, the time window for pollutant migration is accurately determined, and the short-term fluctuation component and the long-term trend component are separated. This overcomes the shortcomings of traditional methods, such as fixed time windows and mixed signals, and improves the accuracy of pollution event identification.

[0012] Third, by constructing a time-delayed causal relationship network and calculating the causal connection strength, the intrinsic relationship between pH, conductivity and dissolved oxygen can be explored. This enables the early detection of potential signs of pollution and the timely generation of alarm records, achieving an improvement from single threshold alarm to multi-parameter logical judgment.

[0013] Fourth, by weighting and correcting the short-term fluctuation components downstream, and comprehensively considering the coupling effect of pollutant decay, water reoxygenation, sediment release and water flow disturbance, the distortion problem of pollution signals in spatial transmission is corrected; further, by integrating the characteristics of tributary inflow and pollutant diffusion, a risk distribution sequence along the river section is generated, high-risk areas are accurately identified and monitoring reports are output, thereby improving the early warning capability and risk control level of water environment monitoring. Attached Figure Description

[0014] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1This is a flowchart of the real-time water environment monitoring method that integrates multi-source sensor data according to this application; Figure 2 This is a risk baseline sequence diagram for this application; Figure 3 This is a comparison chart of the weighted correction effects of this application; Figure 4 This is a comprehensive pollution risk distribution map of the river section in this application; Figure 5 This is a schematic diagram of the structure of the real-time water environment monitoring system that integrates multi-source sensor data according to this application. Figure 6 This is a schematic block diagram of the water environment real-time monitoring device that integrates multi-source sensor data according to this application. Detailed Implementation

[0016] This application provides a method, system, and storage medium for real-time monitoring of the water environment by fusing multi-source sensor data. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0017] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of the real-time water environment monitoring method integrating multi-source sensor data in this application includes: Step S1: Use multi-source sensors to acquire historical records of upstream and downstream monitoring points, process the historical records, and generate a standardized water quality parameter sequence.

[0018] In one specific embodiment, the process of performing step S1 may specifically include the following steps: Historical data on pH, conductivity, and dissolved oxygen content at upstream and downstream monitoring points were obtained using multi-source sensors. Based on the difference in flow velocity caused by the water level difference between upstream and downstream, and the influence of river bend on pollutant retention time, historical data are cleaned and linearly interpolated to generate a standardized water quality parameter sequence.

[0019] Specifically, the multi-source sensors include pH glass electrode sensors, conductivity sensors, and dissolved oxygen fluorescence sensors deployed at monitoring points upstream and downstream of the river. These devices continuously collect and upload historical data at a sampling frequency of once every 15 minutes. Raw historical data containing pH, conductivity, and dissolved oxygen content are obtained from the multi-source sensors. pH reflects the acid-base balance of the water body, conductivity indicates the total concentration of dissolved solids in the water, and dissolved oxygen content characterizes the water body's self-purification capacity and the degree of organic pollution. These three parameters are core indicators of water pollution status, and their changes directly reflect the occurrence and development of pollution events. Considering the actual river environment, the water level difference between upstream and downstream causes flow velocity differences, thus affecting the time it takes for pollutants to migrate downstream. For example, when the upstream water level is 0.5 meters higher than the downstream level, the flow velocity can be estimated to be approximately 0.8 m / s based on the river length. At the same time, the influence of river curvature on pollutant retention time is also significant. For example, in a river section with a curvature (the ratio of the actual river length to the straight-line distance) of 1.5, the pollutant retention time is extended by about 20% compared to a straight river section. These differences in hydrological conditions can cause discrepancies in the timing and morphology of signals generated at upstream and downstream monitoring points for the same pollution event. If the raw data is analyzed directly, it will be difficult to accurately match the pollution correspondence between upstream and downstream.

[0020] To eliminate the impact of these physical factors on data quality, this step incorporates the aforementioned hydrological characteristics during data cleaning and linear interpolation. In the data cleaning stage, the following methods are employed: The principle is to remove outliers, which involves calculating the historical mean and standard deviation of each water quality parameter and removing outliers that exceed the mean. Data points within a range of two standard deviations are marked as outliers and removed. For example, the mean conductivity is... The standard deviation is Then it exceeds Data points within the specified range will be removed. Simultaneously, based on real-time upstream and downstream water level difference data and the actual length of the river segment, the current flow velocity is calculated to assess the theoretical migration time of pollutants. Based on pre-extracted river segment curvature data from the geographic information system, a retention correction coefficient for pollutants in curved river segments is determined to identify periods of data distortion caused by abnormal flow velocity or retention effects. For data gaps due to equipment maintenance or signal loss, linear interpolation is used to fill in the gaps. Specifically, the interpolated value is calculated based on the linear relationship between two adjacent valid data points before and after the missing time. For example, if the missing data is from the morning... If dissolved oxygen data is missing, then according to and as well as and Data trend linear interpolation is obtained The values. After cleaning and interpolation are completed, all parameters are... Standardization is performed to make the mean 0 and the standard deviation 1, thereby eliminating numerical deviations caused by differences in dimensions between different parameters.

[0021] Through the above processing, a standardized water quality parameter sequence is generated. This sequence has complete time domain coverage and a uniform dimensional scale; all parameters have been processed... Standardization, ensuring a mean of 0 and a standard deviation of 1, eliminates numerical biases caused by dimensional differences between parameters, providing an accurate and reliable data foundation for determining pollutant migration time windows. This step, through targeted cleaning and interpolation incorporating hydrological characteristics into the raw monitoring data, effectively addresses data distortion caused by river physical properties, improving data quality and ensuring subsequent analysis accurately reflects water quality conditions, thus laying the groundwork for the accuracy of the entire monitoring method.

[0022] Step S2: Determine the migration time window of pollutants from upstream to downstream based on the standardized water quality parameter sequence.

[0023] In one specific embodiment, the process of performing step S2 may specifically include the following steps: Based on standardized water quality parameter sequences, the time delay of peak concentration arrival time of pH, conductivity and dissolved oxygen content between upstream and downstream monitoring points was calculated using time series cross-correlation functions. Based on the calculation results of the time series cross-correlation function, the similarity matching window of the concentration curves of each water quality parameter is determined; By combining the peak concentration arrival time delay and the similarity matching window, the migration time window of pollutants from upstream to downstream is determined.

[0024] Specifically, based on standardized water quality parameter sequences, the time delays for peak concentrations of pH, conductivity, and dissolved oxygen at upstream and downstream monitoring points are calculated using time-series cross-correlation functions. The time-series cross-correlation function is a statistical method for measuring the similarity between two time series under different time lags. Its calculation logic involves progressively shifting the downstream monitoring point's sequence relative to the upstream monitoring point's sequence, calculating the correlation coefficient for each time unit shift, and identifying the time lag corresponding to the maximum correlation coefficient as the most similar time offset between the two sequences. In this embodiment, the pH sequence at the upstream monitoring point is used as the reference sequence, and the pH sequence at the downstream monitoring point is sequentially delayed by 0 minutes, 15 minutes, 30 minutes, up to 240 minutes. The correlation coefficient is calculated for each lag time, and the lag time corresponding to the maximum correlation coefficient is the time delay for peak concentrations of pH. Taking a specific measurement as an example, the upstream pH level decreased rapidly at 9:00 AM, and the downstream pH level showed a similar decrease at 9:45 AM. The correlation coefficient, calculated using a cross-correlation function, reached 0.92 after a 45-minute lag. This 45-minute lag represents the time delay of the peak pH concentration arrival. Using the same method, the time delays of the peak concentration arrival for conductivity and dissolved oxygen were calculated separately, yielding the individual delay times for each of the three parameters.

[0025] After obtaining the peak concentration arrival time delays for each parameter, the similarity matching window for each water quality parameter concentration curve is further determined based on the calculation results of the time series cross-correlation function. The similarity matching window refers to the time interval during which the upstream and downstream concentration curves exhibit highly similar shapes. It is determined as follows: based on the cross-correlation function calculation results, a correlation coefficient threshold of 0.8 is set. This threshold is determined based on statistical analysis experience of historical monitoring data. The time lag interval where the correlation coefficient continuously exceeds this threshold is extracted as the similarity matching window. The minimum window width is set to 15 minutes. If the width of the interval continuously exceeding the threshold is less than 15 minutes, it is expanded to 15 minutes on both sides of the peak point. Taking pH as an example, on the calculated cross-correlation function curve, the correlation coefficients within the range of 35 to 55 minutes lag all exceed 0.8, and the width is 20 minutes > 15 minutes. Therefore, this 20-minute window is the similarity matching window for pH.

[0026] Based on this, the migration time window of pollutants from upstream to downstream is determined by combining the peak concentration arrival time delay and the similarity matching window. The determination of the migration time window follows these rules: the mean of the peak concentration arrival time delays for each parameter is used as the window center, and the intersection of the similarity matching windows for each parameter is used as the window width, ultimately forming a time interval that simultaneously covers the characteristics of all three parameters. For example, the pH delay is 45 minutes, the conductivity delay is 50 minutes, and the dissolved oxygen delay is 40 minutes, with an average of approximately 45 minutes; the pH matching window is 35 to 55 minutes, the conductivity matching window is 40 to 65 minutes, and the dissolved oxygen matching window is 30 to 50 minutes. The intersection of these three windows is 40 to 50 minutes, thus determining the migration time window as 40 to 50 minutes after the upstream peak occurs. This migration time window serves as the limiting range for subsequent time series extraction, ensuring that subsequent analysis focuses on the effective time interval of pollutant migration from upstream to downstream. This step, through precise calculation of the cross-correlation function of time series and combined with multi-parameter comprehensive judgment, overcomes the limitations of traditional methods that use fixed time windows or single-parameter judgment, and provides an accurate time reference for subsequent time series decomposition and causal analysis.

[0027] Step S3: Based on the migration time window, extract the time series, process the time series using the seasonal trend decomposition method, and obtain the short-term fluctuation component and the long-term trend component.

[0028] In one specific embodiment, the process of performing step S3 may specifically include the following steps: Within the migration time window, time series of pH, conductivity, and dissolved oxygen content were extracted respectively; By processing the time series using the seasonal trend decomposition method, the periodic fluctuations of dissolved oxygen caused by diurnal temperature range and the long-term baseline trends of various water quality parameters were separated. Based on the separation results, short-term volatility components for volatility characteristic analysis and long-term trend components for trend characteristic assessment are generated respectively.

[0029] Specifically, based on the migration time window determined in step S2, continuous time series of pH, conductivity, and dissolved oxygen content are extracted within this window. Taking a certain monitoring as an example, the migration time window is determined to be 40 to 50 minutes after the upstream peak, corresponding to the specific time period from 9:40 to 9:50 AM. Second-level sampling data for each parameter are extracted within this 10-minute interval to form the original time series for subsequent decomposition. These sequence data reflect the dynamic changes of water quality parameters during the migration of pollutants from upstream to downstream. However, they contain signal components of multiple time scales, including periodic fluctuations caused by natural factors such as diurnal alternation and temperature changes, as well as background changes reflecting the long-term evolution trend of water quality. Seasonal trend decomposition methods are needed to effectively separate these components.

[0030] Seasonal trend decomposition is a statistical technique that breaks down a time series into trend components, seasonal components, and residual components. Its processing logic is as follows: The long-term trend of the series is estimated using a moving average filter. Specifically, a moving average is calculated on the original series using a window width equal to the length of the seasonal cycle. For example, for dissolved oxygen data with a 24-hour cycle, a 24-hour moving average window is set to obtain the trend component reflecting the overall trend of the series. Then, the trend component is removed from the original series to obtain a detrended series. Next, the average value at the same position within the cycle is calculated on the detrended series according to the cycle length to extract the seasonal component with a fixed periodicity. Finally, both the trend component and the seasonal component are removed from the original series, leaving the short-term fluctuation component. In this embodiment, for the time series of dissolved oxygen content, this method can effectively separate the periodic fluctuations of dissolved oxygen caused by diurnal temperature variations. Specifically, since the dissolved oxygen content in water is significantly affected by water temperature, the daytime warming due to solar radiation leads to an increase in dissolved oxygen saturation, while the nighttime cooling leads to a decrease in dissolved oxygen, forming a regular fluctuation with a 24-hour cycle. The seasonal trend decomposition method extracts this periodic variation pattern from the dissolved oxygen sequence by setting a period length of 24 hours. For example, it shows a periodic fluctuation curve with a daytime peak of 8.2 mg / L and a nighttime trough of 6.5 mg / L. Simultaneously, this method can also extract the long-term baseline trends of various water quality parameters, reflecting the slow evolution of water quality over several days or weeks. For instance, due to the continuous discharge of industrial wastewater, the long-term trend of conductivity shows a gradual increase, or due to the dilution effect of seasonal rainfall, the long-term trend of pH shows a gradual decrease.

[0031] Based on the separation results above, short-term fluctuation components and long-term trend components are generated. The short-term fluctuation component is the remaining part after removing the trend and seasonal components from the original sequence. It mainly reflects rapid changes caused by sudden pollution events or instantaneous disturbances, such as sudden drops in pH or spikes in conductivity due to illegal industrial wastewater discharge. This component is used for subsequent fluctuation characteristic analysis and abnormal signal identification. The long-term trend component is the trend component extracted using a moving average filter, reflecting the background evolution pattern of water quality. For example, the daily increase in conductivity in rivers due to long-term pollution, or the changes in dissolved oxygen levels caused by the alternation of wet and dry seasons. This component is used for subsequent trend characteristic assessment and baseline comparison. Through the above decomposition process, the original complex water quality parameter sequence is broken down into two components with clear physical meanings. The short-term fluctuation component, used for fluctuation characteristic analysis, focuses on the rapid response characteristics of sudden events, while the long-term trend component, used for trend characteristic assessment, serves the analysis of water quality evolution patterns. This provides high-quality input data for the subsequent construction of a time-delay causal relationship network, enabling the network to focus on the instantaneous response relationship between parameters when pollution events occur, without being disturbed by background trends and periodic noise.

[0032] Step S4: Based on the short-term fluctuation component and the long-term trend component, construct a time-delay causal relationship network to determine the causal connection strength between various water quality parameters.

[0033] In one specific embodiment, the process of performing step S4 may specifically include the following steps: Based on the short-term fluctuation component and the long-term trend component, the synchronous offset time difference between the rapid decrease in pH and the abnormal fluctuation in conductivity is calculated. Based on short-term fluctuation components and long-term trend components, the chain response characteristics of changes in dissolved oxygen content to decreases in pH are analyzed. Based on the synchronization offset time difference and chain response characteristics, a time-delay causal relationship network among various water quality parameters is constructed; By using a time-delay causal correlation network, the causal test statistics between various water quality parameters are calculated, and the causal connection strength between various water quality parameters is determined based on the causal test statistics.

[0034] Specifically, based on short-term fluctuation components and long-term trend components, the synchronous offset time difference between a rapid decrease in pH and an abnormal fluctuation in conductivity is calculated. The synchronous offset time difference refers to the time lag or lead of one parameter relative to another when two parameters undergo abnormal changes, and its calculation is based on the short-term fluctuation components. The synchronous offset time difference is calculated using a sliding window cross-correlation method, with a step size of 15 minutes, searching for the lag time that maximizes the correlation coefficient within the range of 0 to 120 minutes. Taking an industrial wastewater discharge incident as an example, after acidic wastewater was discharged into the river, the short-term pH fluctuation component at the upstream monitoring point showed a rapid decrease of 0.8 pH units at 9:00 AM; simultaneously, the short-term conductivity fluctuation component showed an abnormal increase of 120 microsiemens / cm at 9:12 AM. Through cross-correlation analysis of these two short-term fluctuation component sequences, the lag time of the abnormal conductivity fluctuation relative to the rapid decrease in pH was calculated to be 12 minutes, which is the synchronous offset time difference between the two. This difference reflects the time difference between the acidic components and dissolved solids in the pollutants as they migrate to downstream monitoring points, and is one of the core bases for subsequent judgment of the causal relationship of pollution events.

[0035] Building upon this, we further analyzed the chain response characteristics of dissolved oxygen content changes in response to pH decreases. Chain response characteristics refer to the response pattern where dissolved oxygen content changes following a rapid decrease in pH, including two dimensions: response delay time and response amplitude. Taking the aforementioned industrial wastewater discharge incident as an example, after a rapid decrease in the short-term pH fluctuation component at the upstream monitoring point at 9:00 AM, the short-term dissolved oxygen fluctuation component began a sustained decrease at 9:08 AM, reaching a maximum decrease of 1.5 mg / L at 9:20 AM. Through time-series alignment analysis of the two short-term fluctuation component sequences, we obtained a response delay time of 8 minutes and a response amplitude of 1.5 mg / L for the change in dissolved oxygen content in response to the pH decrease. These two parameters together constitute the chain response characteristics. This characteristic reflects the depletion of dissolved oxygen in water bodies by acidic pollutants and its time scale, and is a key indicator for revealing the ecological effects of pollution events.

[0036] Based on the calculated synchronization offset time difference and chain response characteristics, a time-delay causal relationship network is constructed among various water quality parameters. The time-delay causal relationship network is a directed graph structure with water quality parameters as nodes and causal relationships between parameters as edges, where each edge includes a time delay parameter and causal strength information. In this embodiment, pH, conductivity, and dissolved oxygen content are used as network nodes. Based on the synchronization offset time difference, the directed edge between pH and conductivity is determined to be pH to conductivity, with a time delay of 12 minutes. Based on the chain response characteristics, the directed edge between pH and dissolved oxygen is determined to be pH to dissolved oxygen, with a time delay of 8 minutes. Simultaneously, considering the potential indirect correlation between conductivity and dissolved oxygen, subsequent causal checks further confirm the connection between the two. The network is constructed based on the sudden change information contained in the short-term fluctuation component, while using the long-term trend component as a background reference to ensure that the network reflects the real causal relationship of pollution events rather than pseudo-correlation caused by background noise.

[0037] After the network is constructed, causal test statistics between various water quality parameters are calculated using a time-delay causal association network. The strength of the causal connection between these parameters is then determined based on these statistics. The causal test statistics are calculated using the Granger causality test method. The basic principle is: for a time series system containing two variables, if introducing past values ​​of one variable significantly improves the prediction accuracy of future values ​​of the other variable, then the variable is considered to have a Granger causal relationship with the other variable. In practice, for each pair of directed edges, a restricted model (using only the historical values ​​of the target variable for prediction) and an unrestricted model (incorporating historical values ​​of the source variable for joint prediction) are constructed. The model order for the Granger causality test is determined according to the Akaike Information Criterion, and the F-statistic is calculated based on the sum of squared residuals of the restricted and unrestricted models. The F-statistic is calculated by comparing the sum of squared residuals of the two models; a larger F-value indicates a more significant causal relationship. Taking the causal test of pH on conductivity as an example, the short-term fluctuation component sequences of both parameters are input, and the calculated F-statistic is 8.32, with a significance level p-value less than 0.01, indicating that the decrease in pH has a significant causal effect on abnormal fluctuations in conductivity. This F-statistic is normalized to represent the causal connection strength; for example, normalizing to the 0-1 interval yields 0.87. The causal connection strength is normalized by dividing the F-statistic by the largest F-statistic among all parameter pairs, obtaining values ​​in the 0-1 interval. Following the same method, the causal test statistics for all possible parameter pairs, such as pH on dissolved oxygen and conductivity on dissolved oxygen, are calculated, and connections that pass the significance test are selected to obtain a complete time-delay causal network and the causal connection strength between each parameter. This step, through the construction of the time-delay causal network and the quantification of causal strength, transforms the cooperative change relationship between multiple parameters into a quantifiable causal indicator, providing a scientific basis for the accurate judgment of subsequent abnormal signals and solving the technical problem that traditional methods rely solely on single-parameter threshold alarms and cannot distinguish between causal relationships and accidental correlations.

[0038] Step S5: Based on the time-delay causal relationship network, determine whether there are potential signals of early signs of pollution and generate preliminary abnormal alarm records.

[0039] In one specific embodiment, the process of performing step S5 may specifically include the following steps: Based on the time-delay causal relationship network, it is determined whether the synchronization offset time difference is less than a preset delay threshold and the causal connection strength exceeds a preset strength threshold. If so, it is determined to be a potential signal of early signs of pollution. Based on the judgment results, a preliminary abnormal alarm record is generated.

[0040] Specifically, based on the constructed time-delay causal relationship network, the calculated synchronization offset time difference and causal connection strength are obtained from the network. The synchronization offset time difference reflects the temporal relationship between the rapid decrease in pH and the abnormal fluctuation in conductivity. For example, in the aforementioned embodiment, the lag time of the abnormal conductivity fluctuation relative to the rapid decrease in pH is 12 minutes. The causal connection strength quantifies the predictive ability of pH changes on conductivity changes; for example, the strength value obtained after Granger causality test normalization is 0.87. These two indicators together constitute the core basis for determining whether a pollution event has occurred.

[0041] Based on this, it is determined whether the synchronization offset time difference is less than a preset delay threshold and the causal connection strength exceeds a preset strength threshold. (Preset delay threshold) Based on the hydrological characteristics of the river section and statistical analysis of historical monitoring data, the specific value is determined to be the mean of the synchronous offset time difference under normal historical conditions plus twice the standard deviation, but not exceeding 30 minutes. For example, historical tracing experiments on this river section show that under normal hydrological conditions, the time difference for different components of pollutants to reach the downstream area is usually between 15 and 25 minutes, with a mean of 20 minutes and a standard deviation of 2.5 minutes. 20 plus 2 multiplied by 2.5 equals 25 minutes, meaning that a synchronization offset time difference of less than 25 minutes indicates a high temporal correlation between changes in pH and conductivity. (Preset intensity threshold) Based on the significance level setting of the Granger causality test, the critical value of the F-statistic corresponding to a p-value less than 0.05 is taken. As a criterion for judgment. The value of depends on the model order p and the sample size n, and can be obtained by looking up the F-distribution table. For example, when p equals 2 and n equals 100... The value is approximately 3.09. The normalized strength threshold is determined based on the distribution of the F-statistic of historical data. For example, the 90th percentile of the historical F-statistic is used as the strength threshold, typically ranging from 0.65 to 0.85. In this embodiment, the synchronization offset time difference is 12 minutes, which is less than the preset delay threshold of 25 minutes; the normalized value of the causal connection strength is 0.87, which exceeds the preset strength threshold of 0.75 determined based on historical data statistics. Both conditions are met simultaneously.

[0042] When both of the above conditions are met, it is determined to be a potential signal of early pollution. This determination means that the rapid decrease in pH and the abnormal fluctuation in conductivity are not accidental simultaneous occurrences, but rather have a clear temporal correlation and statistical causal relationship, and are highly likely to be caused by the same pollution event. For example, the illegal discharge of acidic industrial wastewater causes a decrease in the pH value of the water body, while the dissolved solids carried in the wastewater cause an increase in conductivity. The two maintain a relatively stable time difference during the migration process, and this correlation is statistically significant, therefore it is determined to be an early sign of pollution. This determination mechanism effectively distinguishes between genuine pollution events and false signals caused by sensor noise, natural fluctuations, or accidental superposition of different pollution sources, significantly improving the accuracy of early warning.

[0043] Based on the judgment results, a preliminary anomaly alarm record is generated. This record contains the following key information: judgment time, upstream monitoring point location, downstream monitoring point location, relevant pollution parameters (pH, conductivity, dissolved oxygen), synchronization offset time difference (12 minutes), causal connection strength (0.87), preset delay threshold (25 minutes), preset intensity threshold (0.75), and the judgment conclusion of early pollution symptoms. The preliminary anomaly alarm record is stored in the system database in a structured data format and marked as the original alarm information to be processed by subsequent weighted correction. This record serves as an intermediate bridge from signal identification to signal correction in the entire monitoring method, providing clear input basis for subsequent steps. It ensures that only reliable signals verified by causal logic can enter the weighted correction stage, avoiding interference from invalid alarms to subsequent analysis and improving the overall operational efficiency and early warning reliability of the monitoring method.

[0044] Step S6: Based on the preliminary abnormal alarm records, perform weighted correction on the short-term fluctuation components, and identify the impact effect based on the weighted correction results.

[0045] In one specific embodiment, the process of performing step S6 may specifically include the following steps: Based on the preliminary abnormal alarm records, and combined with the pollutant concentration attenuation coefficient along the path and the correlation characteristics of water reoxygenation rate and aeration intensity, the short-term fluctuation components of downstream monitoring points are analyzed. Based on the analysis results, the short-term fluctuation components are weighted and corrected. Based on the weighted correction results, the influence of the coupling effect between pollutants released from sediment and water flow disturbance on the pollution signal was identified.

[0046] Specifically, based on preliminary anomaly alarm records, information on upstream pollution events indicated in these records and the corresponding short-term fluctuation components at downstream monitoring points were obtained. The preliminary anomaly alarm records confirmed a significant time-lag causal relationship between the rapid decrease in pH and abnormal fluctuations in conductivity, indicating that a pollution event did indeed occur upstream, and the pollution signal is migrating downstream. However, the pollution signal does not simply maintain its original form and intensity during its transmission along the river; rather, it is affected by the combined effects of multiple environmental factors, including the natural decay of pollutants, the reoxygenation capacity of the water body, and the coupling influence of riverbed sediment and water flow disturbance. Therefore, further analysis of the short-term fluctuation components at downstream monitoring points is needed to correct for signal distortion caused by these factors.

[0047] In the specific analysis, the pollutant concentration attenuation coefficient along the migration path and the correlation characteristics between the water reoxygenation rate and the aeration intensity are considered. The pollutant concentration attenuation coefficient along the migration path refers to the rate at which the concentration of pollutants gradually decreases as the migration distance increases. Its value is related to factors such as the nature of the pollutants, water temperature, and riverbed adsorption characteristics. For example, for degradable organic pollutants, the attenuation coefficient is usually between 0.05 / km and 0.15 / km. The water reoxygenation rate refers to the speed at which the water body replenishes dissolved oxygen through the exchange between the water surface and the atmosphere. The aeration intensity reflects the enhancing effect of water flow disturbance on the reoxygenation process. There is a positive correlation between the two, that is, the more turbulent the water flow and the greater the aeration intensity, the higher the water reoxygenation rate. In this embodiment, based on the parameters of the historical water quality model of the river section, the pollutant concentration attenuation coefficient along the migration path of the current river section is set to 0.08 / km, and the correlation coefficient between the water reoxygenation rate and the aeration intensity is set to 0.6, that is, the actual reoxygenation rate is 1.6 times the theoretical reoxygenation rate of a calm water surface. By combining these parameters with the pollution event information carried in the preliminary anomaly alarm records, a quantitative analysis of the short-term fluctuation components of downstream monitoring points is performed to assess the theoretical intensity variation range that the pollution signal should have when it reaches the downstream.

[0048] Based on the analysis results, a weighted correction is applied to the short-term fluctuation components. The core logic of the weighted correction is as follows: based on the attenuation and reoxygenation parameters determined in the above analysis, a weighted correction function is constructed to adjust the short-term fluctuation component sequence of downstream monitoring points point by point. The correction process first normalizes the original short-term fluctuation components, unifying their amplitudes to the range of 0 to 1, eliminating dimensional differences between different parameters; then, based on the theoretical migration time of the pollution signal and the river length, the attenuation compensation coefficient and reoxygenation influence coefficient corresponding to each time point are calculated; finally, the normalized fluctuation components are multiplied by the comprehensive weighting coefficient to obtain the corrected short-term fluctuation components. Taking the aforementioned embodiment as an example, after the upstream pollution event, the original amplitude of the short-term conductivity fluctuation component detected by the downstream monitoring point within the predicted arrival time window is 0.72 (after normalization). Based on the attenuation coefficient of 0.08 / km and the river length of 2km, the theoretical attenuation amplitude is 16%, meaning the original signal should be compensated to 0.83; simultaneously, based on the measured aeration intensity, the reoxygenation effect causes a slight dilution of the conductivity signal, and the correction coefficient is taken as 0.95. The overall weighting coefficient is composed of the attenuation compensation coefficient and the reoxygenation correction coefficient, i.e., 0.83 × 0.95 = 0.79. The final weighted correction for the short-term conductivity fluctuation component is 0.79. The same method is used to weight and correct the short-term fluctuation components of pH and dissolved oxygen separately.

[0049] After weighted correction, the coupling effect of sediment-released pollutants and water flow disturbance on the pollution signal is identified based on the weighted correction results. Sediment release refers to the process by which riverbed sediments are resuspended and release adsorbed pollutants under the influence of water flow disturbance. This process may lead to abnormal amplification of the pollution signal at downstream monitoring points. Water flow disturbance, on the other hand, affects the dilution and diffusion of pollutants by changing the mixing state of the water body. The coupling effect of these two factors may cause the intensity of the pollution signal detected at downstream monitoring points to deviate from the theoretical value predicted based on the attenuation model. In this embodiment, the weighted correction of the short-term fluctuation component is compared with the theoretical prediction value. If the corrected amplitude is significantly higher than the theoretical prediction value (e.g., more than 20%), it is determined that there is an amplification effect of sediment release; if the corrected amplitude is significantly lower than the theoretical prediction value, it is determined that there is a dilution and attenuation effect of water flow disturbance; if the two are basically consistent, it indicates that the coupling effect is not significant. Taking the aforementioned conductivity correction result as an example, the weighted correction amplitude is 0.79, while the theoretical prediction value considering only attenuation and reoxygenation is 0.83, which is about 5% lower than the actual value. This indicates that the dilution effect caused by water flow disturbance is slightly dominant, and the sediment release effect is not significant. This effect serves as an important input parameter for subsequent risk assessment, characterizing the actual changes in the pollution signal due to local environmental factors during spatial transmission. This step effectively corrects the distortion problem of the pollution signal during transmission by weighting and correcting the short-term fluctuation components of downstream monitoring points and identifying coupling effects. This allows subsequent risk assessments to be based on data that more closely reflects the actual situation, significantly improving the adaptability of the monitoring method to complex river environments and the accuracy of risk assessment.

[0050] Step S7: Based on the weighted and corrected short-term fluctuation components and long-term trend components, generate a risk baseline sequence, and further combine the impact effects to generate a comprehensive pollution risk distribution sequence. Based on the comprehensive pollution risk distribution sequence, form a monitoring work guidance report.

[0051] In one specific embodiment, the process of performing step S7 may specifically include the following steps: Obtain the weighted adjusted short-term volatility component and long-term trend component, as well as the impact effect; The weighted and adjusted short-term volatility component is superimposed with the long-term trend component to generate a risk baseline sequence; Based on the risk baseline sequence, the comprehensive pollution risk distribution sequence along the river section is generated by integrating the characteristics of conductivity abrupt changes caused by upstream and downstream tributary confluence points, the characteristics of concentration curve broadening time intervals caused by pollutant diffusion, and the impact effects. Based on the comprehensive pollution risk distribution sequence, identify and determine the priority monitoring scope for high-risk areas; Based on the priority monitoring areas, a monitoring work guidance report is generated.

[0052] Specifically, step S6 obtains the weighted and corrected short-term fluctuation components and long-term trend components, as well as the identified impact effects. The weighted and corrected short-term fluctuation components have been compensated for by pollutant attenuation and corrected for reoxygenation effects, and can truly reflect the actual impact intensity of the pollution event at downstream monitoring points; the long-term trend components characterize the background evolution pattern of water quality in this river section, such as the seasonal increase in conductivity or the diurnal baseline variation of dissolved oxygen; the impact effects record the degree of amplification or attenuation of the pollution signal by the coupling effect of sediment release and water flow disturbance, and are important correction factors for subsequent risk assessment.

[0053] Based on this, the weighted adjusted short-term fluctuation components are superimposed with the long-term trend components to generate a risk baseline sequence. The superposition process uses a linear superposition method, that is, the weighted adjusted short-term fluctuation component value and the long-term trend component value at the same moment are added to form a new sequence reflecting the overall water quality status at that moment. Taking a certain river section as an example, the weighted adjusted short-term conductivity fluctuation component at 10:00 AM is 0.79 (normalized value), and the corresponding long-term trend component is 0.45. After superposition, the risk baseline value at that moment is 1.24. This risk baseline sequence integrates the instantaneous impact of sudden pollution events and the long-term evolution of water quality, comprehensively reflecting the actual water quality risk level at each monitoring moment, and providing a basic carrier for subsequent risk feature fusion.

[0054] Based on the risk baseline sequence, the comprehensive pollution risk distribution sequence along the river segment is generated by further integrating the conductivity abrupt change characteristics caused by upstream and downstream tributary confluence points, the concentration curve broadening time interval characteristics caused by pollutant diffusion, and the impact effect. The conductivity abrupt change characteristic refers to the phenomenon of a step change in conductivity at the tributary confluence point due to the mixing of different water bodies. For example, if a tributary carries high-salinity wastewater into the main stream, the conductivity downstream of the confluence point may instantly increase by 30%. The location and magnitude of this abrupt change are incorporated into the risk baseline sequence as spatial characteristic parameters. The concentration curve broadening time interval characteristics refer to the phenomenon of prolonged concentration peak duration caused by diffusion during pollutant migration. For example, after pollutants instantaneously discharged upstream reach downstream, their concentration curve broadens from the original 5-minute width to 15 minutes. This broadening coefficient is used to adjust the distribution pattern of the risk baseline sequence in the time dimension. The impact effect serves as an amplitude correction factor, adjusting the values ​​of the corresponding river segment in the risk baseline sequence. The fusion process employs a weighted fusion algorithm, which comprehensively calculates the risk contribution weights of each feature parameter. The weight coefficients are determined based on statistical analysis of historical pollution event data. For example, the weight for conductivity mutation is 0.3, the weight for broadening is 0.2, and the weight for impact effect is 0.5. Finally, a comprehensive risk score is generated for each river section, forming a comprehensive pollution risk distribution sequence that is continuously distributed along the river from upstream to downstream.

[0055] Based on the comprehensive pollution risk distribution sequence, priority monitoring areas for high-risk regions are identified and determined. The identification method employs a threshold determination approach, setting a risk score threshold of, for example, 0.8. Areas within a continuous river segment with a risk score exceeding 0.8 are marked as high-risk areas. Taking a specific monitoring session as an example, the comprehensive pollution risk distribution sequence shows that from 2.5 km downstream of the upstream monitoring point to 4.8 km downstream, the risk score continuously exceeds 0.8, peaking at 1.35. Therefore, this 2.3 km long river segment is identified as the priority monitoring area for high-risk regions. This range reflects the location and length of the river segment most severely affected by the pollution event and serves as the core basis for subsequent monitoring resource allocation.

[0056] Based on the priority monitoring area, a monitoring work guidance report is generated. This report is output in a structured document format and includes the following key information: report generation time, start and end locations of the priority monitoring area (e.g., 2.5 km to 4.8 km downstream of the upstream monitoring point), length of the high-risk river section (2.3 km), main contributing parameters (abrupt changes in conductivity, abnormal pH), recommended monitoring frequency (e.g., increasing from the usual once every 4 hours to once every 30 minutes), coordinates of recommended mobile monitoring equipment deployment points, and potential pollution sources requiring close attention (e.g., an industrial discharge outlet within the river section). The monitoring work guidance report can be directly distributed to on-site monitoring personnel or automated dispatch systems to guide subsequent intensive monitoring, sampling analysis, and pollution source tracing. This step, through risk baseline construction, multi-feature fusion, risk area identification, and monitoring report generation, transforms the analysis results of previous steps into actionable monitoring guidance information, achieving a complete closed loop from data acquisition, signal processing, causal analysis to risk management, significantly improving the early warning capability and emergency response efficiency of water environment monitoring.

[0057] refer to Figure 2 This figure is a risk baseline sequence diagram, showing the risk baseline sequence generated by superimposing the weighted adjusted short-term fluctuation component and the long-term trend component, focusing on the period of pollution event occurrence (days 2.5-3.5). It effectively separates signals from different time scales that are mixed together in traditional methods, and then re-superimposes them to form the risk baseline. This allows monitoring personnel to capture both the rapid changes (short-term fluctuations) of sudden pollution events and grasp the background evolution patterns of water quality (long-term trends), overcoming the shortcomings of mixed signals in traditional methods.

[0058] refer to Figure 3 This figure shows a comparison of the weighted correction effect, comparing the differences between the original upstream signal, the original downstream signal, and the corrected downstream signal. This corresponds to the weighted correction process. By comparing the signal differences before and after correction, it can be clearly seen that this method effectively solves the technical problem of "downstream signal distortion" in traditional monitoring. The figure shows that the corrected green curve is closer to the actual pollution situation.

[0059] refer to Figure 4 This map shows the comprehensive pollution risk distribution along the river, displaying the sequence of comprehensive pollution risks along the river channel, marking the locations of tributary confluence points, and identifying high-risk areas with red semi-transparent areas. This corresponds to the process of generating the comprehensive pollution risk distribution sequence and identifying high-risk areas in step S7. Figure 4 The precise marking of the semi-transparent red area is a direct demonstration of this effect. Traditional methods can only provide point-like information indicating an alarm at a certain monitoring point, while... Figure 4 It can output continuous spatial guidance information such as "key monitoring is needed from X kilometers to Y kilometers", enabling monitoring resources to be accurately deployed to the most needed locations, significantly improving the efficiency and accuracy of risk management.

[0060] It is understood that the executing entity of this application can be a real-time water environment monitoring system that integrates multi-source sensor data, or it can be a terminal or a server; the specific implementation is not limited here. This application's embodiment uses a server as an example for illustration.

[0061] The above describes the real-time water environment monitoring method integrating multi-source sensor data in the embodiments of this application. The following describes the real-time water environment monitoring system integrating multi-source sensor data in the embodiments of this application. Please refer to [link / reference]. Figure 5 One embodiment of the real-time water environment monitoring system integrating multi-source sensor data in this application includes: The data processing module is used to acquire historical records of upstream and downstream monitoring points using multi-source sensors, process the historical records, and generate standardized water quality parameter sequences. The window definition module is used to determine the migration time window of pollutants from upstream to downstream based on a standardized water quality parameter sequence; The time series decomposition module is used to extract time series based on migration time windows, process the time series through seasonal trend decomposition methods, and obtain short-term fluctuation components and long-term trend components. The network construction module is used to construct a time-delay causal relationship network based on short-term fluctuation components and long-term trend components, and to determine the causal connection strength between various water quality parameters. The signal recognition module is used to determine whether there are potential signals of early signs of pollution based on time-delay causal correlation networks, and generate preliminary abnormal alarm records. The weighted correction module is used to perform weighted correction on the short-term fluctuation components based on the preliminary abnormal alarm records, and to identify the impact effect based on the weighted correction results. The risk assessment module is used to generate a risk baseline sequence based on the weighted and corrected short-term fluctuation components and long-term trend components. It further combines the impact effects to generate a comprehensive pollution risk distribution sequence, and forms a monitoring work guidance report based on the comprehensive pollution risk distribution sequence.

[0062] Through the collaborative efforts of the aforementioned components, the data processing module first cleans and standardizes the historical records of upstream and downstream monitoring points, generating a standardized water quality parameter sequence that provides a unified data foundation for subsequent analysis. The window definition module calculates the pollutant migration time window based on this sequence, ensuring that subsequent analysis focuses on the effective data range. The time series decomposition module extracts the time series within the migration time window and performs seasonal trend decomposition, breaking down the complex water quality signal into short-term fluctuation components reflecting sudden pollution and long-term trend components reflecting long-term evolution patterns. The network construction module constructs a time-delay causal relationship network based on these two components, and by calculating the causal connection strength between various parameters, it mines the correlations between pH, conductivity, and dissolved oxygen. The inherent correlation between the data sources and the signal recognition module relies on this network to comprehensively judge whether the synchronization offset time difference and causal connection strength meet the preset conditions, capturing potential signs in the early stages of pollution and generating alarm records. The weighted correction module, based on the alarm records and combining the pollutant attenuation characteristics along the river and water reoxygenation factors, corrects the downstream short-term fluctuation components, identifies the coupling effect of sediment release and water flow disturbance on the pollution signal, and effectively corrects the distortion problem of pollution signals in spatial transmission. The risk assessment module finally superimposes the corrected components with the long-term trend components to form a risk baseline sequence, and integrates tributary inflow mutations and pollutant diffusion characteristics to generate a comprehensive pollution risk distribution sequence along the river section, outputting high-risk areas and guidance for monitoring work. The entire scheme progresses step-by-step from data input to result output, effectively solving problems such as ignoring hydrological differences, insufficient mining of time-delay relationships, and signal transmission distortion in existing technologies.

[0063] above Figure 5 The real-time water environment monitoring system integrating multi-source sensor data in this embodiment of the invention is described in detail from the perspective of modular functional entities. The real-time water environment monitoring device integrating multi-source sensor data in this embodiment of the invention is described in detail from the perspective of hardware processing.

[0064] Reference Figure 6 This invention also provides a real-time water environment monitoring device that integrates multi-source sensor data. This device can be a server, and its internal structure can be as follows: Figure 6As shown, the real-time water environment monitoring device integrating multi-source sensor data includes a processor, memory, display screen, input device, network interface, and database connected via a system bus. The processor in this computer design provides computing and control capabilities. The memory of the real-time water environment monitoring device integrating multi-source sensor data includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database of the real-time water environment monitoring device integrating multi-source sensor data is used to store the data corresponding to this embodiment. The network interface of the real-time water environment monitoring device integrating multi-source sensor data is used for communication with external terminals via network connection. When the computer program is executed by the processor, it can implement the above-described method.

[0065] Those skilled in the art will understand that Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present invention and does not constitute a limitation on the water environment real-time monitoring device that integrates multi-source sensor data and is applied thereto.

[0066] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when the instructions are executed on a computer, cause the computer to perform the steps of the method for real-time monitoring of the water environment by fusing multi-source sensor data.

[0067] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0068] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0069] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for real-time monitoring of the water environment by integrating multi-source sensor data, characterized in that, The method includes: S1. Use multi-source sensors to acquire historical records of upstream and downstream monitoring points, process the historical records, and generate a standardized water quality parameter sequence. S2. Based on the standardized water quality parameter sequence, determine the migration time window of pollutants from upstream to downstream; S3. Based on the migration time window, extract the time series, process the time series using the seasonal trend decomposition method, and obtain the short-term fluctuation component and the long-term trend component. S4. Based on the short-term fluctuation component and the long-term trend component, construct a time-delay causal relationship network to determine the causal connection strength between each water quality parameter; S5. Based on the time-delay causal relationship network, determine whether there are potential signals of early signs of pollution and generate preliminary abnormal alarm records; S6. Based on the preliminary abnormal alarm record, perform weighted correction on the short-term fluctuation component, and identify the impact effect based on the weighted correction result; S7. Based on the weighted and corrected short-term fluctuation component and the long-term trend component, a risk baseline sequence is generated. Further, combined with the impact effect, a comprehensive pollution risk distribution sequence is generated. Based on the comprehensive pollution risk distribution sequence, a monitoring work guidance report is formed.

2. The method according to claim 1, characterized in that, S1 includes: Historical data on pH, conductivity, and dissolved oxygen content at upstream and downstream monitoring points were obtained using multi-source sensors. Based on the difference in flow velocity caused by the water level difference between upstream and downstream, and the influence of river bend on pollutant retention time, the historical records are cleaned and linearly interpolated to generate a standardized water quality parameter sequence.

3. The method according to claim 1, characterized in that, S2 includes: Based on the standardized water quality parameter sequence, the peak concentration arrival time delay of pH, conductivity and dissolved oxygen content between the upstream and downstream monitoring points was calculated using the time series cross-correlation function. Based on the calculation results of the time series cross-correlation function, the similarity matching window of the concentration curves of each water quality parameter is determined; By combining the peak concentration arrival time delay and the similarity matching window, the migration time window of pollutants from upstream to downstream is determined.

4. The method according to claim 3, characterized in that, S3 includes: Within the migration time window, time series of pH, conductivity, and dissolved oxygen content were extracted respectively. The time series was processed using a seasonal trend decomposition method to separate the periodic fluctuations of dissolved oxygen caused by diurnal temperature differences from the long-term baseline trends of various water quality parameters. Based on the separation results, short-term volatility components for volatility characteristic analysis and long-term trend components for trend characteristic assessment are generated respectively.

5. The method according to claim 1, characterized in that, S4 includes: Based on the short-term fluctuation component and the long-term trend component, calculate the synchronous offset time difference between the rapid decrease in pH and the abnormal fluctuation in conductivity. Based on the short-term fluctuation component and the long-term trend component, the chain response characteristics of the change in dissolved oxygen content to the decrease in pH are analyzed. Based on the synchronization offset time difference and the chain response characteristics, a time-delay causal relationship network among various water quality parameters is constructed. The causal test statistics between water quality parameters are calculated using the time-delay causal correlation network, and the causal connection strength between water quality parameters is determined based on the causal test statistics.

6. The method according to claim 5, characterized in that, S5 includes: Based on the time-delay causal relationship network, it is determined whether the synchronization offset time difference is less than a preset delay threshold and the causal connection strength exceeds a preset strength threshold. If so, it is determined to be a potential signal of early signs of pollution. Based on the judgment results, a preliminary abnormal alarm record is generated.

7. The method according to claim 6, characterized in that, S6 includes: Based on the preliminary abnormal alarm records, and combined with the pollutant concentration decay coefficient along the path and the correlation characteristics of water reoxygenation rate and aeration intensity, the short-term fluctuation components of the downstream monitoring points are analyzed. Based on the analysis results, the short-term fluctuation components are weighted and corrected. Based on the weighted correction results, the influence of the coupling effect between pollutants released from sediment and water flow disturbance on the pollution signal was identified.

8. The method according to claim 7, characterized in that, S7 includes: Obtain the weighted adjusted short-term volatility component and the long-term trend component, as well as the impact effect; The weighted and corrected short-term volatility component is superimposed with the long-term trend component to generate a risk baseline sequence; Based on the aforementioned risk baseline sequence, the conductivity abrupt change characteristics caused by upstream and downstream tributary confluence points, the concentration curve broadening time interval characteristics caused by pollutant diffusion, and the aforementioned impact effects are integrated to generate a comprehensive pollution risk distribution sequence along the river section. Based on the comprehensive pollution risk distribution sequence, identify and determine the priority monitoring scope for high-risk areas; Based on the aforementioned priority monitoring scope, a monitoring work guidance report is generated.

9. A real-time water environment monitoring system integrating multi-source sensor data, used to implement the method as described in any one of claims 1-8, characterized in that, The real-time water environment monitoring system that integrates multi-source sensor data includes: The data processing module is used to acquire historical records of upstream and downstream monitoring points using multi-source sensors, process the historical records, and generate standardized water quality parameter sequences. The window definition module is used to determine the migration time window of pollutants from upstream to downstream based on a standardized water quality parameter sequence; The time series decomposition module is used to extract time series based on migration time windows, process the time series through seasonal trend decomposition methods, and obtain short-term fluctuation components and long-term trend components. The network construction module is used to construct a time-delay causal relationship network based on short-term fluctuation components and long-term trend components, and to determine the causal connection strength between various water quality parameters. The signal recognition module is used to determine whether there are potential signals of early signs of pollution based on time-delay causal correlation networks, and generate preliminary abnormal alarm records. The weighted correction module is used to perform weighted correction on the short-term fluctuation components based on the preliminary abnormal alarm records, and to identify the impact effect based on the weighted correction results. The risk assessment module is used to generate a risk baseline sequence based on the weighted and corrected short-term fluctuation components and long-term trend components. It further combines the impact effects to generate a comprehensive pollution risk distribution sequence, and forms a monitoring work guidance report based on the comprehensive pollution risk distribution sequence.

10. A computer-readable storage medium storing instructions thereon, characterized in that, When the instruction is executed by the processor, it implements the water environment real-time monitoring method that integrates multi-source sensor data as described in any one of claims 1-8.

Images