Water quality abnormity early warning system and method based on multi-source data fusion

Abstract

The invention provides a water quality abnormity early warning system and method based on multi-source data fusion, and relates to the technical field of abnormity early warning, and the method comprises the steps: recognizing a sensitive region in which the water quality is gradually changed and abnormal in a target water area through the time-space correlation characteristics of each piece of water quality multi-source data; performing sensitivity analysis on the response intensity of the water quality gradual change abnormity in the sensitive area to obtain a water quality sensitivity index of the sensitive area; determining an early warning response period when the water quality at the monitoring point is gradually changed and abnormal; performing confidence correction on a water quality abnormal fluctuation early warning threshold value at the monitoring point through the water quality sensitivity index and the early warning response period to obtain an abnormal early warning confidence coefficient of the water quality at the monitoring point; and when the abnormity early warning confidence exceeds a preset confidence threshold, sending a water quality gradual change abnormity verification instruction to an intelligent sensor at the monitoring point, and generating risk alarm information of water quality gradual change abnormity based on feedback verification data. According to the invention, risk early warning can be carried out under the condition that the water quality has gradual abnormal fluctuation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of abnormal early warning, more specifically, the present application relates to a water quality abnormal early warning system and method based on multi-source data fusion. BACKGROUND

[0002] Abnormal early warning is a method based on data collection, analysis and model inference, which is used to identify potential risks or abnormal states in the process of system operation. The core idea is to extract abnormal features and judge their evolution trend by real-time perception of key monitoring parameters, combined with statistical modeling, signal processing, machine learning or artificial intelligence algorithms, so as to give early warning signals. This technology is widely used in environmental monitoring, equipment operation and maintenance, medical health and financial risk control, etc. Compared with traditional post-diagnosis, abnormal early warning emphasizes forward-looking and real-time, which can intervene before the abnormality develops into failure or disaster, and provides reliable basis for risk management and decision support.

[0003] With the acceleration of industrialization and urbanization, the types and discharge modes of pollutants in water environment are becoming increasingly complex, and the water quality changes gradually show multi-source, multi-scale and nonlinear characteristics. Traditional water quality monitoring and early warning methods mostly rely on single-point data or static threshold judgment, which has certain identification ability in sudden pollution events, but it is difficult to capture and accurately predict the gradual abnormal fluctuations caused by the gradual accumulation, migration and diffusion of pollutants in water body. Especially under the joint action of natural hydrological process, pollution source discharge load and water self-purification capacity, the change of water quality parameters usually shows a dynamic process of slow evolution and stage mutation. The complex evolution law makes the early warning strategy of single index or fixed threshold lack of adaptability, and it is easy to cause early warning lag or misjudgment. Therefore, how to carry out risk early warning under the condition of gradual abnormal fluctuations of water quality has become a key problem to be solved in the field of water environment management and pollution prevention and control. SUMMARY

[0004] The present application provides a water quality abnormal early warning system and method based on multi-source data fusion, which can realize risk early warning under the condition of gradual abnormal fluctuations of water quality.

[0005] In the first aspect, the present application provides a water quality abnormal early warning method based on multi-source data fusion, which comprises the following steps: Collecting water quality multi-source data based on intelligent sensors at different monitoring points of the target water area, and identifying sensitive areas of gradual abnormal water quality in the target water area through the spatio-temporal correlation characteristics of each water quality multi-source data; performing sensitivity analysis on the response intensity of the water quality gradual abnormality in the sensitive area based on the cumulative load characteristics of each pollution source in a monitoring period, to obtain a water quality sensitivity index of the sensitive area; determining a warning response period of the water quality gradual abnormality at the monitoring point according to the deviation degree of the water quality change trend at the monitoring point and the seasonal fluctuation of the water quality change; performing confidence correction on the water quality abnormal fluctuation warning threshold at the monitoring point through the water quality sensitivity index and the warning response period, to obtain an abnormal warning confidence of the water quality at the monitoring point; when the abnormal warning confidence exceeds a preset confidence threshold, sending a water quality gradual abnormality verification instruction to the intelligent sensor at the monitoring point, and generating a risk alarm information of the water quality gradual abnormality based on the feedback verification data.

[0006] In the embodiment, the sensitive area in which the water quality in the target water area presents a gradual abnormality is identified through the spatio-temporal correlation characteristics of each water quality multi-source data, and specifically includes: preprocessing each water quality multi-source data, and then determining the spatio-temporal correlation characteristics of each water quality multi-source data after preprocessing; determining a reference spatio-temporal correlation characteristic based on historical normal water quality monitoring data of the target water area; comparing the reference spatio-temporal correlation characteristic with all spatio-temporal correlation characteristics, and screening out monitoring points whose spatio-temporal correlation characteristics exceed the normal fluctuation interval of the reference spatio-temporal correlation characteristic; performing spatial clustering analysis on the screened monitoring points, and then obtaining the sensitive area in which the water quality in the target water area presents a gradual abnormality.

[0007] In the embodiment, the sensitivity analysis on the response intensity of the water quality gradual abnormality in the sensitive area based on the cumulative load characteristics of each pollution source in a monitoring period, to obtain the water quality sensitivity index of the sensitive area specifically includes: obtaining pollution source information of the sensitive area in a monitoring period, the pollution source information including pollution source type, discharge location, discharge frequency, pollutant type and initial discharge concentration; determining the cumulative load characteristics of each pollution source in the monitoring period according to the pollutant input amount of each pollution source to the target water area in each time period in the pollution source information; determining a hydrological correlation degree between each pollution source and the sensitive area based on the spatial distance, hydrological connectivity and water retention time of each pollution source and the sensitive area; determining an abnormal sensitivity coefficient of each pollution source to the response intensity of the water quality gradual abnormality in the sensitive area according to the cumulative load characteristics of each pollution source in the monitoring period and the hydrological correlation degree between each pollution source and the sensitive area; statistically determining the gradual abnormality amplitude of the water quality index at each monitoring point in the sensitive area; The water quality sensitive index of the sensitive area is determined according to all of the abnormal sensitive coefficients and the gradual abnormal amplitude of the water quality indexes at the monitoring points.

[0008] In the embodiment, the early warning response period of the gradual abnormality of the water quality at the monitoring point is determined according to the deviation degree of the water quality change trend at the monitoring point and the seasonal fluctuation of the water quality change, and specifically includes the following steps: determining the deviation degree of the water quality change trend at the monitoring point; determining the seasonal fluctuation of the water quality change at the monitoring point; dividing the early warning response level according to the deviation degree of the water quality change trend at the monitoring point; determining the period fluctuation of the gradual abnormality of the water quality at the monitoring point according to the seasonal fluctuation of the water quality change at the monitoring point; determining the early warning response period of the gradual abnormality of the water quality at the monitoring point according to the early warning response level and the period fluctuation.

[0009] In the embodiment, the abnormal early warning confidence of the water quality at the monitoring point is obtained by confidence correction of the water quality abnormal fluctuation early warning threshold at the monitoring point according to the water quality sensitive index and the early warning response period, and specifically includes the following steps: obtaining the water quality abnormal fluctuation early warning threshold at the monitoring point; correcting the water quality abnormal fluctuation early warning threshold at the monitoring point according to the water quality sensitive index to obtain a corrected water quality abnormal fluctuation early warning threshold; confidence evaluating the corrected water quality abnormal fluctuation early warning threshold according to the early warning response period to obtain the abnormal early warning confidence of the water quality at the monitoring point.

[0010] In the embodiment, the water quality gradual abnormality verification instruction is sent to the intelligent sensor at the monitoring point, and the risk alarm information of the water quality gradual abnormality is generated based on the feedback verification data, and specifically includes the following steps: sending the water quality gradual abnormality verification instruction to the intelligent sensor at the monitoring point, and receiving the water quality verification data feedback by the intelligent sensor; abnormal evaluating the water quality verification data to obtain the abnormal evaluation result of the water quality at the monitoring point; generating the risk alarm information of the water quality gradual abnormality according to the abnormal evaluation result.

[0011] In the embodiment, the water quality sensitive index represents the overall sensitive degree of the sensitive area to the water quality gradual abnormality.

[0012] In the embodiment, the early warning response period represents the period of the early warning response to the gradual abnormality of the water quality at the monitoring point.

[0013] In the embodiment, the abnormal early warning confidence degree represents a confidence degree of early warning of the gradual abnormal fluctuation of the water quality at the monitoring point.

[0014] In a second aspect, the application provides a water quality abnormal early warning system based on multi-source data fusion, which is used to execute a water quality abnormal early warning method based on multi-source data fusion. The water quality abnormal early warning system comprises: A sensitive area identification module is configured to identify a sensitive area in which the water quality presents a gradual abnormality in the target water area based on the water quality multi-source data collected by the intelligent sensors at different monitoring points of the target water area and the spatio-temporal correlation features of the water quality multi-source data. A sensitivity analysis module is configured to analyze the response intensity of the gradual abnormality of the water quality in the sensitive area based on the cumulative load characteristics of the pollution sources in the monitoring period, and obtain a water quality sensitivity index of the sensitive area. An early warning response period determination module is configured to determine an early warning response period of the gradual abnormality of the water quality at the monitoring point according to the deviation degree of the water quality change trend and the seasonal fluctuation of the water quality change. A confidence correction module is configured to correct the water quality abnormal fluctuation early warning threshold at the monitoring point based on the water quality sensitivity index and the early warning response period, and obtain an abnormal early warning confidence degree of the water quality at the monitoring point. An early warning information generation module is configured to send a water quality gradual abnormality verification instruction to the intelligent sensor at the monitoring point when the abnormal early warning confidence degree exceeds a preset confidence threshold, and generate a risk alarm information of the water quality gradual abnormality based on the feedback verification data. The technical scheme provided by the embodiments of the application has the following beneficial effects: The sensitive area in which the water quality presents a gradual abnormality in the target water area is identified based on the water quality multi-source data collected by the intelligent sensors at different monitoring points of the target water area and the spatio-temporal correlation features of the water quality multi-source data. The response intensity of the gradual abnormality of the water quality in the sensitive area is analyzed based on the cumulative load characteristics of the pollution sources in the monitoring period, and a water quality sensitivity index of the sensitive area is obtained. An early warning response period of the gradual abnormality of the water quality at the monitoring point is determined according to the deviation degree of the water quality change trend and the seasonal fluctuation of the water quality change. The water quality abnormal fluctuation early warning threshold at the monitoring point is corrected based on the water quality sensitivity index and the early warning response period, and an abnormal early warning confidence degree of the water quality at the monitoring point is obtained. When the abnormal early warning confidence degree exceeds a preset confidence threshold, a water quality gradual abnormality verification instruction is sent to the intelligent sensor at the monitoring point, and a risk alarm information of the water quality gradual abnormality is generated based on the feedback verification data.

[0015] It can be seen that in the present application, when the abnormal early warning confidence exceeds the preset confidence threshold, a water quality gradual change abnormality verification instruction is sent to the intelligent sensor at the monitoring point, and risk warning information of water quality gradual change abnormality is generated based on the feedback verification data; wherein, first, after obtaining water quality multi-source data of different monitoring points in the target water area, the sensitive area of water quality gradual change abnormality is identified through the correlation characteristics between the multi-source data, realizing dynamic identification from "point" to "area", which can more comprehensively reveal the accumulation and diffusion process of pollutants in water body; secondly, the sensitivity analysis mechanism based on the cumulative load characteristics of pollution sources is introduced, the long-term cumulative effect of pollution sources and the response strength of water quality change are combined, and the water quality sensitivity index is constructed, which not only quantifies the response difference of different regions to pollution load, but also effectively reflects the coupling relationship between the progressive influence of pollutants and the self-purification ability of water body, thereby improving the sensitivity of identifying water quality gradual change abnormality risk; thirdly, further considering the trend deviation and seasonal fluctuation factors of water quality parameters, the early warning response period of gradual change abnormality is determined through joint analysis of the water quality change trend and seasonal cycle at the monitoring point, overcoming the shortcomings of traditional methods ignoring long-term evolution law, so that the early warning result can dynamically adapt to the time characteristics of water quality change, avoiding false alarms caused by seasonal disturbance; then, the water quality sensitivity index and the early warning response period are combined to correct the confidence of the abnormal fluctuation warning threshold, and the water quality abnormal early warning confidence at the monitoring point is obtained, realizing the adaptive adjustment of the early warning threshold, which not only improves the scientificity and rationality of the early warning threshold, but also reduces the false alarm and missed alarm risk caused by improper setting of the static threshold; finally, when the abnormal early warning confidence exceeds the preset threshold, the system automatically issues a verification instruction to the intelligent sensor, and generates risk warning information based on the feedback data, thereby forming a closed-loop mechanism of "early warning-verification-confirmation", which not only improves the accuracy and reliability of risk warning, but also avoids the deviation caused by single algorithm judgment.

[0016] In summary, the technical scheme adopted by the present application can realize risk early warning under the condition that water quality presents gradual abnormal fluctuation. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical scheme in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are only the embodiments of the present application, and other drawings can be obtained by those skilled in the art without creating labor intensity.

[0018] Figure 1 is a flow chart of a water quality abnormal early warning method based on multi-source data fusion provided by the present application; Figure 2is a flowchart of a process for determining a sensitive area according to the present application; Figure 3 is a flowchart of a process for determining an abnormal early warning confidence according to the present application; Figure 4 is a module structure diagram of a water quality abnormal early warning system based on multi-source data fusion according to the present application. DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0020] The embodiments of the present application provide a water quality abnormal early warning system and method based on multi-source data fusion. The core is that water quality multi-source data is collected by intelligent sensors at different monitoring points of a target water area, a sensitive area in which water quality presents gradual abnormality is identified through the spatio-temporal correlation characteristics of each water quality multi-source data, the response intensity of the gradual abnormality of water quality in the sensitive area is analyzed based on the cumulative load characteristics of each pollution source in a monitoring period, a water quality sensitivity index of the sensitive area is obtained, the early warning response period of the gradual abnormality of water quality at the monitoring point is determined according to the deviation of the water quality change trend and the seasonal fluctuation of the water quality change, the water quality abnormal fluctuation early warning threshold at the monitoring point is confidence corrected through the water quality sensitivity index and the early warning response period, an abnormal early warning confidence of the water quality at the monitoring point is obtained, when the abnormal early warning confidence exceeds a preset confidence threshold, a water quality gradual abnormality verification instruction is sent to the intelligent sensor at the monitoring point, and risk alarm information of the water quality gradual abnormality is generated based on the feedback verification data.

[0021] In order to better understand the above technical solutions, the above technical solutions will be described in detail below in combination with the drawings in the specification and specific implementation manners. Referring to FIG. 1, the figure is an exemplary flowchart of a water quality abnormal early warning method based on multi-source data fusion according to the embodiments of the present application. The water quality abnormal early warning method comprises the following steps: Figure 1 In step S1, water quality multi-source data is collected by intelligent sensors at different monitoring points of a target water area, and a sensitive area in which water quality presents gradual abnormality is identified through the spatio-temporal correlation characteristics of each water quality multi-source data.

[0022] ​In a specific implementation, the water quality multi-source data collected by the intelligent sensors at different monitoring points in the target water area can be implemented in the following manner: the water quality multi-source data is collected by the intelligent sensors arranged at different monitoring points in the target water area, wherein the water quality multi-source data is composed of data of basic physical indexes, key chemical indexes, and ecological indexes at different time points, wherein the basic physical indexes specifically include water temperature, turbidity, water flow rate, suspended solid concentration, and conductivity; the key chemical indexes specifically include pH value, dissolved oxygen content, nitrate nitrogen content, sulfide content, ammonia nitrogen content, and total phosphorus content; and the ecological indexes specifically include chlorophyll a content, cyanobacterial cell concentration, and phycocyanin content.

[0023] It should be noted that the intelligent sensor in the present application refers to a water quality monitoring device integrating data collection, analysis, and communication functions, which can automatically detect and dynamically perceive various physical, chemical, and ecological parameters of the target water area, and realize real-time perception and intelligent identification of the water quality state.

[0024] Preferably, in the present embodiment, the sensitive area in which the water quality in the target water area presents gradual abnormality is identified through the spatiotemporal correlation characteristics of each water quality multi-source data, and the sensitive area is determined according to the following formula: Figure 2 As shown in the figure, the figure is a flowchart for determining the sensitive area in some embodiments of the present application, and the sensitive area in the present embodiment can be implemented in the following steps: In step S11, each water quality multi-source data is preprocessed, and then the spatiotemporal correlation characteristics of each water quality multi-source data after preprocessing are determined; In step S12, the reference spatiotemporal correlation characteristics are determined based on the historical normal water quality monitoring data of the target water area; In step S13, the reference spatiotemporal correlation characteristics are compared with all spatiotemporal correlation characteristics, and the monitoring points whose spatiotemporal correlation characteristics exceed the normal fluctuation range of the reference spatiotemporal correlation characteristics are screened out; In step S14, spatial clustering analysis is performed on all screened monitoring points, and then the sensitive area in which the water quality in the target water area presents gradual abnormality is obtained.

[0025] It should be noted that the spatiotemporal correlation characteristics in the present application represent the characteristics of the continuous evolution law of the water quality index in time and the mutual influence relationship in space; the reference spatiotemporal correlation characteristics represent the spatiotemporal correlation reference standard when the water quality is in a healthy state; and the sensitive area represents the concentrated distribution area in which the water quality in the target water area presents gradual abnormality.

[0026] In a specific implementation, first, the noise data in each water quality multi-source data is removed by using the Python Pandas library (for example, the abnormal values are identified and deleted by using the 3σ criterion), then the missing data is completed by linear interpolation, and then the index dimension is unified by using the Min-Max standardization. Then, for each pre-processed water quality multi-source data, the trend slope and the coefficient of variation of each index in the pre-processed water quality multi-source data in the sliding window are extracted by using the sliding window technology, and then the Pearson correlation coefficient of each index in the pre-processed water quality multi-source data between adjacent monitoring points in the sliding window is calculated. The matrix composed of all the trend slopes, the coefficients of variation, and the Pearson correlation coefficients in the time sequence is used as the spatiotemporal correlation feature of the pre-processed water quality multi-source data. Second, the historical normal water quality monitoring data of the target water area in the past 3-5 years (the data of the period of known pollution events needs to be removed) is obtained, and then the spatiotemporal correlation feature of the normal water quality monitoring data is calculated by using the Python Pandas library, and the obtained spatiotemporal correlation feature is used as the reference spatiotemporal correlation feature. Further, the normal fluctuation interval of the reference spatiotemporal correlation feature is calculated (for example, the mean value ± 2 times the standard deviation of the coefficient of variation, and the obtained interval is used as the normal interval of the coefficient of variation in the reference spatiotemporal correlation feature), and the interval composed of the normal intervals of the coefficients of variation, the trend slopes, and the Pearson correlation coefficients in the reference spatiotemporal correlation feature is used as the normal fluctuation interval of the reference spatiotemporal correlation feature. Then, all the spatiotemporal correlation features are compared with the reference spatiotemporal correlation feature one by one, and the monitoring points corresponding to the coefficients of variation, the trend slopes, and the Pearson correlation coefficients in the spatiotemporal correlation features that exceed the normal fluctuation interval of the reference spatiotemporal correlation feature are determined. Finally, the DBSCAN clustering algorithm is used to cluster all the monitoring points by taking the longitude and latitude of the monitoring points as the spatial coordinates and the mean value of the Pearson correlation coefficients in the spatiotemporal correlation features as the clustering basis, and the region composed of the largest monitoring point cluster obtained after clustering is used as the sensitive region in which the water quality in the target water area presents a gradual abnormality.

[0027] In step S2, the response intensity of the gradual abnormality of the water quality in the sensitive region is analyzed based on the cumulative load characteristics of each pollution source in the monitoring period to obtain the water quality sensitivity index of the sensitive region.

[0028] In this embodiment, the response intensity of the gradual abnormality of the water quality in the sensitive region is analyzed based on the cumulative load characteristics of each pollution source in the monitoring period to obtain the water quality sensitivity index of the sensitive region, which can be realized by the following steps: The pollution source information of the sensitive region in the monitoring period is obtained, and the pollution source information includes the pollution source type, the discharge location, the discharge frequency, the pollutant type, and the initial discharge concentration. The cumulative load characteristics of each pollution source in the monitoring period are determined according to the pollutant input amount of each pollution source in each time period in the pollution source information. determining a hydrological correlation degree between each pollution source and the sensitive area based on spatial distance, hydrological connectivity and water body retention time between each pollution source and the sensitive area; determining an abnormal sensitivity coefficient of each pollution source to abnormal gradual change response strength of water quality in the sensitive area according to cumulative load characteristics of each pollution source in a monitoring period and the hydrological correlation degree between each pollution source and the sensitive area; counting gradual change abnormal amplitudes of different water quality indexes at each monitoring point in the sensitive area; determining a water quality sensitivity index of the sensitive area according to all abnormal sensitivity coefficients and the gradual change abnormal amplitudes of water quality indexes at each monitoring point.

[0029] It should be noted that the cumulative load characteristics in the present application represent the total amount of pollutants input by pollution sources to water area, reflecting the strength of the influence of pollution sources on water quality; the hydrological correlation degree represents the correlation strength of measuring the influence accessibility of pollution sources and sensitive areas; the abnormal sensitivity coefficient represents the sensitive degree of pollution sources to the potential influence of water quality gradual change abnormality in the sensitive area, and the gradual change abnormal amplitude represents the response degree of water quality indexes to water quality gradual change abnormality; and the water quality sensitivity index represents the overall sensitive degree of the sensitive area to water quality gradual change abnormality.

[0030] In a specific implementation, first, the pollution source information of the sensitive area in a monitoring period (for example, one quarter) is obtained, the monitoring period is divided into several time periods according to weeks, the formula "single time period pollutant input amount = initial emission concentration x emission flow x emission frequency" is used, the single pollutant (for example, total phosphorus) input amount of each pollution source is calculated by the Python Pandas library hour by hour, the total input amount of the corresponding pollution source in the monitoring period is obtained by accumulation, and then all the total input amounts are normalized, and the value obtained after the normalization is taken as the cumulative load characteristic of the corresponding pollution source in the monitoring period; second, the spatial distance from the pollution source discharge port to the sensitive area is calculated by GPS, the SWAT hydrological model is used to simulate the river network of the basin, the hydrological connectivity between the pollution source and the sensitive area is determined (for example, the pollution source is located upstream of the sensitive area and is not blocked by a dam, the connectivity coefficient is set to 1, and is set to 0.2 if it is located downstream), and then the value obtained by weighting the spatial distance (the weight is set to 0.3), the hydrological connectivity (the weight is set to 0.4), and the residence time (the weight is set to 0.3) is taken as the hydrological correlation degree between the pollution source and the sensitive area, so as to obtain the hydrological correlation degree between each pollution source and the sensitive area; then, the cumulative load characteristic of each pollution source is multiplied by the hydrological correlation degree between each pollution source and the sensitive area, and the value obtained by the multiplication is taken as the abnormal sensitivity coefficient of the corresponding pollution source to the water quality gradual abnormal response strength in the sensitive area, so as to obtain the abnormal sensitivity coefficient of each pollution source to the water quality gradual abnormal response strength in the sensitive area; then, the average value of each water quality index in the monitoring period is extracted from the preprocessed water quality monitoring data of each monitoring point in the sensitive area, and compared with the average value of the historical normal water quality index of the sensitive area in the past 3-5 years, and the gradual abnormal amplitude of each water quality index is obtained by the formula: gradual abnormal amplitude = (monitoring water quality index average value-normal water quality index average value) / normal water quality index average value; finally, the abnormal sensitivity coefficient of each pollution source to the water quality gradual abnormal response strength in the sensitive area is multiplied by the gradual abnormal amplitude of different water quality indexes at each monitoring point in the sensitive area, and then the value obtained by the multiplication is normalized, and the average value of the value obtained after the normalization is further calculated, and the average value is taken as the water quality sensitivity index of the sensitive area.

[0031] In step S3, the pre-warning response period of the gradual abnormality of the water quality at the monitoring point is determined according to the deviation of the water quality change trend at the monitoring point and the seasonal fluctuation of the water quality change.

[0032] In this embodiment, the pre-warning response period of the gradual abnormality of the water quality at the monitoring point can be determined according to the deviation of the water quality change trend at the monitoring point and the seasonal fluctuation of the water quality change by the following steps: The deviation of the water quality change trend at the monitoring point is determined. determining seasonal fluctuation of water quality change at the monitoring point; dividing early warning response levels according to deviation degree of water quality change trend at the monitoring point; determining periodic fluctuation amount of gradual abnormality of water quality at the monitoring point according to seasonal fluctuation of water quality change at the monitoring point; determining early warning response period of gradual abnormality of water quality at the monitoring point according to the early warning response levels and the periodic fluctuation amount.

[0033] It should be noted that the deviation degree in the present application represents the degree of deviation of the monitored water quality trend from the long-term normal trend; the seasonal fluctuation represents the characteristic of seasonal fluctuation of the water quality index; the early warning response level represents the urgency level of early warning; the periodic fluctuation amount represents the dynamic change amount of water quality fluctuation between different seasons; and the early warning response period represents the period of early warning response to gradual abnormality of water quality at the monitoring point.

[0034] In a specific implementation, first, based on the "pure natural state" water quality monitoring data of the monitoring point in the past 3-5 years without pollution events and extreme weather, the historical normal trend of different water quality indicators is extracted using the STL time series decomposition method of the Python statsmodels library (for example, the stable trend slope of total phosphorus in the same period each year is 0.01 mg / L / month, representing the normal decline of natural degradation), then linear regression is performed on the water quality monitoring data of different water quality indicators in the monitoring period to obtain the actual trend of different water quality indicators, then the absolute value of the difference between the slope of the actual trend of different water quality indicators and the slope of the historical normal trend of the corresponding water quality indicator is taken, and then the mean of all absolute values is taken as the deviation degree of the water quality change trend at the monitoring point; secondly, based on the water quality monitoring data of the monitoring point in the past 3-5 years, the water quality monitoring data is divided according to different seasons to obtain water quality monitoring data segments in different seasons, then the standard deviation of different water quality indicators in all water quality monitoring data segments is calculated, and the set of all standard deviations is taken as the seasonal fluctuation of water quality change at the monitoring point; then, if the deviation degree of the water quality change trend at the monitoring point is less than 10%, the warning response level is set to a low warning level, if the deviation degree of the water quality change trend at the monitoring point is 10%-30%, the warning response level is set to a medium warning level, and if the deviation degree of the water quality change trend at the monitoring point is greater than 30%, the warning response level is set to a high warning level; then, all standard deviations in the seasonal fluctuation of water quality change at the monitoring point are sorted in chronological order, then difference calculation is performed on the obtained sequence, then each value obtained by difference calculation is divided by the corresponding seasonal time width, then the sum of all values obtained by division is calculated, and the value obtained by summing is taken as the periodic fluctuation amount of the water quality at the monitoring point presenting gradual abnormality; finally, if the warning response level is a high warning level and the periodic fluctuation amount is greater than 0.2, the warning response period of the water quality at the monitoring point presenting gradual abnormality is set to 1 day / time, if the warning response level is a high warning level and the periodic fluctuation amount is between 0.05 and 0.2, the warning response period of the water quality at the monitoring point presenting gradual abnormality is set to 2 days / time, if the warning response level is a high warning level and the periodic fluctuation amount is less than 0.05, the warning response period of the water quality at the monitoring point presenting gradual abnormality is set to 3 days / time, if the warning response level is a medium warning level and the periodic fluctuation amount is greater than 0.2, the warning response period of the water quality at the monitoring point presenting gradual abnormality is set to 5 days / time, and if the warning response level is a low warning level, the warning response period of the water quality at the monitoring point presenting gradual abnormality is 7 days / time.

[0035] In step S4, the water quality anomaly fluctuation warning threshold at the monitoring point is confidence corrected by the water quality sensitive index and the warning response period to obtain the abnormal warning confidence of the water quality at the monitoring point.

[0036] Preferably, in the present embodiment, the water quality anomaly fluctuation early warning threshold at the monitoring point is confidence corrected by the water quality sensitivity index and the early warning response period, to obtain an anomaly early warning confidence of the water quality at the monitoring point, and the anomaly early warning confidence is referenced Figure 3 As shown in the figure, the figure is a flowchart for determining the anomaly early warning confidence in some embodiments of the present application, and the anomaly early warning confidence in the present embodiment can be achieved by the following steps: In step S41, the water quality anomaly fluctuation early warning threshold at the monitoring point is obtained. In step S42, the water quality anomaly fluctuation early warning threshold at the monitoring point is corrected according to the water quality sensitivity index, to obtain a corrected water quality anomaly fluctuation early warning threshold. In step S43, the corrected water quality anomaly fluctuation early warning threshold is confidence evaluated by the early warning response period, to obtain the anomaly early warning confidence of the water quality at the monitoring point.

[0037] It should be noted that the water quality anomaly fluctuation early warning threshold in the present application represents the early warning critical value of the water quality anomaly fluctuation at the monitoring point, which can be directly obtained from the intelligent sensor at the monitoring point. It should be noted that the anomaly early warning confidence in the present application represents the confidence degree of the early warning of the gradual anomaly fluctuation of the water quality at the monitoring point.

[0038] In specific implementation, first, if the water quality sensitivity index belongs to low sensitivity, i.e. the water quality sensitivity index is 0-0.3, the correction coefficient is set to 1.0, if the water quality sensitivity index belongs to medium sensitivity, i.e. the water quality sensitivity index is 0.31-0.70, the correction coefficient is set to 0.85, and if the water quality sensitivity index belongs to high sensitivity, i.e. the water quality sensitivity index is 0.71-1, the correction coefficient is set to 0.7, the water quality anomaly fluctuation early warning threshold is multiplied by the corresponding correction coefficient, and the multiplied value is taken as the corrected water quality anomaly fluctuation early warning threshold; second, the quarterly data amount corresponding to the early warning response period is calculated by Python pandas (for example, 90 corresponding to 1 day / time), and then the 95% confidence interval is calculated by scipy.stats.t.interval, and the probability that the confidence interval covers the corrected water quality anomaly fluctuation early warning threshold is taken as the anomaly early warning confidence of the water quality at the monitoring point.

[0039] In step S5, when the anomaly early warning confidence exceeds a preset confidence threshold, a water quality gradual anomaly verification instruction is sent to the intelligent sensor at the monitoring point, and risk alarm information of the water quality gradual anomaly is generated based on the feedback verification data.

[0040] It should be noted that the mean value of the anomaly early warning confidence of the water quality at all monitoring points in the present application can be taken as the preset confidence threshold.

[0041] In the embodiment, the water quality gradual change abnormality verification instruction is sent to the intelligent sensor at the monitoring point, and the risk alarm information of the water quality gradual change abnormality can be generated based on the feedback verification data by using the following steps: The water quality gradual change abnormality verification instruction is sent to the intelligent sensor at the monitoring point, and the water quality verification data feedback by the intelligent sensor is received; The water quality verification data is abnormally evaluated to obtain the abnormal evaluation result of the water quality at the monitoring point; The risk alarm information of the water quality gradual change abnormality is generated according to the abnormal evaluation result.

[0042] In a specific implementation, first, a water quality gradual change abnormality verification instruction is sent to the intelligent sensor at the monitoring point, wherein the water quality gradual change abnormality verification instruction needs to specify the key parameters: one is the verification index list (for example, the conductivity, chemical oxygen demand, pH value, total phosphorus content, and chlorophyll a content for this verification), and the other is the sampling frequency and time length (for example, collecting data once every 5 minutes for 30 minutes), after the intelligent sensor receives the water quality gradual change abnormality verification instruction, the water quality verification data is collected and packaged and fed back to the Internet of Things platform; then, the Internet of Things platform first pre-processes the received water quality verification data: using the sliding window method to eliminate obvious outliers (for example, the chemical oxygen demand suddenly increases to 100 mg / L and then instantaneously recovers, which is determined as sensor fault noise), and then using the mean value before and after to fill in the short-time missing data (for example, the 15th minute data is not transmitted back, and the values of the 10th minute and the 20th minute are used to estimate), and then performing abnormality evaluation: first step, single-index analysis, comparing the measured values of each index with the preset gradual change abnormality threshold (for example, the conductivity threshold is 500 μS / cm, and the total phosphorus threshold is 0.15 mg / L), and marking the threshold exceeding time; at the same time, the change rate in 30 minutes is calculated by linear fitting (for example, the total phosphorus increases from 0.1 mg / L to 0.2 mg / L, and the change rate is 100%), if the change rate exceeds 50% and the threshold exceeds 10 minutes or more, it is determined that the index is “gradual change abnormality”, the second step is multi-index comprehensive judgment, through the platform rule engine correlation analysis: if only the pH value slightly exceeds the threshold and the change rate is low, it is determined as “low risk”; if the total phosphorus (change rate 80%) and chlorophyll a (change rate 120%) exceed the threshold at the same time, forming a coherent abnormality of “nutrient salt accumulation→algal reproduction”, then it is determined as “medium risk”, if the conductivity, chemical oxygen demand, and cyanobacterial cell concentration all significantly exceed the threshold and the change rate exceeds 100%, it is determined as “high risk”, and thus the abnormality evaluation result of the water quality at the monitoring point is obtained; finally, based on the abnormality evaluation result, the key contents are integrated in a fixed format, that is, the specific location of the monitoring point, the abnormal index and abnormal value (for example, total phosphorus 0.2 mg / L, exceeding threshold 33%), risk level (for example, medium risk), gradual change trend description (rising continuously in the last 30 minutes), and suggested measures, through the message pushing function of the Internet of Things platform, the information is sent to the environmental protection monitoring personnel in the form of short message and APP pop-up window, thereby forming the risk alarm information of water quality gradual change abnormality.

[0043] It can be seen that in the present application, when the abnormal early warning confidence exceeds the preset confidence threshold, a water quality gradual change abnormality verification instruction is sent to the intelligent sensor at the monitoring point, and risk warning information of water quality gradual change abnormality is generated based on the feedback verification data; wherein, first, after obtaining water quality multi-source data of different monitoring points in the target water area, the sensitive area of water quality gradual change abnormality is identified through the correlation characteristics between the multi-source data, realizing dynamic identification from "point" to "area", which can more comprehensively reveal the accumulation and diffusion process of pollutants in water body; secondly, the sensitivity analysis mechanism based on the cumulative load characteristics of the pollution source is introduced, the long-term cumulative effect of the pollution source and the response strength of water quality change are combined, and the water quality sensitivity index is constructed, which not only quantifies the response difference of different regions to pollution load, but also effectively reflects the coupling relationship between the progressive influence of pollutants and the self-purification capacity of water body, thereby improving the sensitivity of identifying the risk of water quality gradual change abnormality; thirdly, further considering the trend deviation and seasonal fluctuation factors of water quality parameters, the early warning response period of gradual change abnormality is determined through joint analysis of the water quality change trend and seasonal cycle at the monitoring point, overcoming the shortcomings of traditional methods ignoring long-term evolution law, so that the early warning result can dynamically adapt to the time characteristics of water quality change, avoiding false alarms caused by seasonal disturbance; then, the water quality sensitivity index and the early warning response period are combined to correct the confidence of the abnormal fluctuation warning threshold, and the water quality abnormal early warning confidence at the monitoring point is obtained, realizing the adaptive adjustment of the early warning threshold, which not only improves the scientificity and rationality of the early warning threshold, but also reduces the false alarm and missed alarm risk caused by improper setting of the static threshold; finally, when the abnormal early warning confidence exceeds the preset threshold, the system automatically issues a verification instruction to the intelligent sensor, and generates risk warning information based on the feedback data, thereby forming a closed-loop mechanism of "early warning-verification-confirmation", which not only improves the accuracy and reliability of risk warning, but also avoids the deviation caused by single algorithm judgment.

[0044] In summary, the technical scheme adopted by the present application can realize risk early warning under the condition that water quality presents gradual abnormal fluctuation.

[0045] In embodiment two, the present application provides a water quality abnormal early warning system based on multi-source data fusion, as shown in Figure 4 The water quality abnormal early warning system includes: A sensitive area identification module 100 is used to collect water quality multi-source data based on intelligent sensors at different monitoring points in the target water area, and identify the sensitive area of water quality gradual change abnormality in the target water area through the spatio-temporal correlation characteristics of each water quality multi-source data; The sensitivity analysis module 200 is configured to perform sensitivity analysis on the response intensity of the water quality gradual change anomaly in the sensitive area based on the cumulative load characteristics of each pollution source in the monitoring period, and obtain a water quality sensitivity index of the sensitive area; The early warning response period determination module 300 is configured to determine the early warning response period of the water quality gradual change anomaly at the monitoring point according to the deviation degree of the water quality change trend at the monitoring point and the seasonal fluctuation of the water quality change. The confidence correction module 400 is configured to perform confidence correction on the water quality anomaly fluctuation early warning threshold at the monitoring point by using the water quality sensitivity index and the early warning response period, and obtain an anomaly early warning confidence of the water quality at the monitoring point. The early warning information generation module 500 is configured to send a water quality gradual change anomaly verification instruction to the intelligent sensor at the monitoring point when the anomaly early warning confidence exceeds a preset confidence threshold, and generate risk alarm information of the water quality gradual change anomaly based on the feedback verification data.

[0046] The present application is described with reference to flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The device that implements the functions specified in one flow or multiple flows and / or blocks Figure 1 The device that implements the functions specified in one flow or multiple flows and / or blocks

[0047] Those skilled in the art can understand that all or part of the steps of various methods in the above embodiments can be completed by instructing the relevant hardware by means of a program, and the program can be stored in a computer readable storage medium, including Read-Only Memory (ROM), Random Access Memory (RAM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), One-time Programmable Read-Only Memory (OTPROM), Electrically-Erasable Programmable Read-Only Memory (EEPROM), Compact Disc Read-Only Memory (CD-ROM) or other optical disk storage, magnetic disk storage, magnetic tape storage, or any other medium that can be used to carry or store data in a computer readable manner.

[0048] It should also be noted that the terms "comprising", "containing", or any other variant thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements, but can also include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the statement "comprising a" does not exclude the existence of additional identical elements in the process, method, article, or apparatus that includes the element.

Claims

1. A water quality anomaly early warning method based on multi-source data fusion, characterized in that, The water quality anomaly early warning method comprises the following steps: The water quality anomaly early warning method comprises the following steps: Based on the intelligent sensor at different monitoring points of the target water area, the water quality multi-source data is collected, and the sensitive area where the water quality in the target water area presents gradual abnormality is identified through the space-time correlation characteristics of each water quality multi-source data; Based on the cumulative load characteristics of each pollution source in the monitoring period, the response intensity of the gradual abnormality of the water quality in the sensitive area is analyzed for sensitivity, and the water quality sensitivity index of the sensitive area is obtained; According to the deviation of the water quality change trend at the monitoring point and the seasonal fluctuation of the water quality change, the early warning response period of the gradual abnormality of the water quality at the monitoring point is determined; The water quality anomaly early warning method comprises the following steps: 2.The water quality anomaly early warning method based on multi-source data fusion of claim 1, wherein, When the abnormal early warning confidence exceeds the preset confidence threshold, the water quality gradual abnormality verification instruction is sent to the intelligent sensor at the monitoring point, and the risk alarm information of the water quality gradual abnormality is generated based on the feedback verification data. The sensitive area where the water quality in the target water area presents gradual abnormality is identified through the space-time correlation characteristics of each water quality multi-source data, which comprises: The space-time correlation characteristics of each water quality multi-source data after preprocessing are determined; The baseline space-time correlation characteristics are determined based on the historical normal water quality monitoring data of the target water area; The baseline space-time correlation characteristics are compared with all space-time correlation characteristics, and the monitoring points whose space-time correlation characteristics exceed the normal fluctuation interval of the baseline space-time correlation characteristics are screened out; 3.The water quality anomaly early warning method based on multi-source data fusion of claim 1, wherein, The spatial clustering analysis is performed on the screened monitoring points, and then the sensitive area where the water quality in the target water area presents gradual abnormality is obtained. Based on the cumulative load characteristics of each pollution source in the monitoring period, the response intensity of the gradual abnormality of the water quality in the sensitive area is analyzed for sensitivity, and the water quality sensitivity index of the sensitive area is obtained, which comprises: The pollution source information of the sensitive area in the monitoring period is obtained, and the pollution source information comprises the pollution source type, the discharge position, the discharge frequency, the pollutant type and the initial discharge concentration; The cumulative load characteristics of each pollution source in the monitoring period are determined according to the pollutant input amount of each pollution source to the target water area in each time period in the pollution source information; The hydrological correlation degree between each pollution source and the sensitive area is determined based on the spatial distance, the hydrological connectivity and the water retention time of each pollution source and the sensitive area; The abnormal sensitivity coefficient of each pollution source to the response intensity of the gradual abnormality of the water quality in the sensitive area is determined according to the cumulative load characteristics of each pollution source in the monitoring period and the hydrological correlation degree between each pollution source and the sensitive area; The gradual abnormality amplitude of the water quality index at each monitoring point in the sensitive area is counted; 4. The water quality anomaly early warning method based on multi-source data fusion according to claim 1, characterized in that, The water quality sensitivity index of the sensitive area is determined according to all abnormal sensitivity coefficients and the gradual abnormality amplitude of the water quality index at each monitoring point. The early warning response period of the gradual abnormality of the water quality at the monitoring point is determined according to the deviation of the water quality change trend at the monitoring point and the seasonal fluctuation of the water quality change, which comprises: The deviation of the water quality change trend at the monitoring point is determined; The seasonal fluctuation of the water quality change at the monitoring point is determined; The early warning response level is determined based on the degree of deviation of the water quality change trend at the monitoring point; The periodic fluctuation of water quality at the monitoring point is determined based on the seasonal fluctuation of water quality changes at the monitoring point. The warning response period for a gradual change in water quality at a monitoring point is determined based on the warning response level and the periodic fluctuation amount.

5. The water quality anomaly early warning method based on multi-source data fusion according to claim 1, characterized in that, The confidence level of the water quality anomaly warning threshold at the monitoring point is obtained by performing confidence correction on the water quality sensitivity index and the warning response period, specifically including: Obtain the early warning threshold for abnormal water quality fluctuations at monitoring points; The water quality abnormal fluctuation warning threshold at the monitoring point is corrected according to the water quality sensitivity index to obtain the corrected water quality abnormal fluctuation warning threshold. The confidence level of the abnormal water quality warning at the monitoring point is obtained by evaluating the corrected water quality abnormal fluctuation warning threshold through the warning response cycle.

6. The water quality anomaly early warning method based on multi-source data fusion according to claim 1, characterized in that, Sending water quality gradual change anomaly verification commands to the smart sensors at the monitoring points, and generating risk alarm information for water quality gradual change anomalies based on the feedback verification data, specifically includes: Send water quality gradual change anomaly verification instructions to the smart sensors at the monitoring points, and receive water quality verification data fed back by the smart sensors; Anomaly assessment was performed on the water quality verification data to obtain anomaly assessment results for the water quality at the monitoring points; Based on the anomaly assessment results, risk alarm information for gradual water quality anomalies is generated.

7. The water quality anomaly early warning method based on multi-source data fusion according to claim 1, characterized in that, The water quality sensitivity index indicates the overall sensitivity of a sensitive area to gradual changes in water quality. 8.The water quality anomaly early warning method based on multi-source data fusion of claim 1, wherein, The early warning response period refers to the period during which an early warning response is issued when the water quality at the monitoring point shows a gradual change in abnormality. 9.The water quality anomaly early warning method based on multi-source data fusion of claim 1, wherein, The anomaly warning confidence level indicates the degree of reliability in issuing warnings about gradual abnormal fluctuations in water quality at monitoring points.

10. A water quality anomaly early warning system based on multi-source data fusion, configured to perform a water quality anomaly early warning method based on multi-source data fusion according to any one of claims 1 to 9. The water quality anomaly early warning system includes: The sensitive area identification module is used to collect multi-source water quality data from intelligent sensors at different monitoring points in the target water area, and to identify sensitive areas in the target water area where the water quality shows gradual abnormalities through the spatiotemporal correlation characteristics of the multi-source water quality data. The sensitivity analysis module is used to perform sensitivity analysis on the response intensity of gradual water quality anomalies in the sensitive area based on the cumulative load characteristics of each pollution source within the monitoring period, and to obtain the water quality sensitivity index of the sensitive area. The early warning response cycle determination module is used to determine the early warning response cycle when the water quality at the monitoring point shows a gradual abnormality based on the deviation of the water quality change trend and the seasonal fluctuation of water quality changes at the monitoring point. The confidence correction module is used to perform confidence correction on the water quality abnormal fluctuation warning threshold at the monitoring point using the water quality sensitivity index and the warning response cycle, so as to obtain the abnormal warning confidence level of the water quality at the monitoring point. The early warning information generation module is used to send a water quality gradual change anomaly verification command to the smart sensor at the monitoring point when the confidence level of the abnormal early warning exceeds a preset confidence threshold, and generate a risk alarm information for water quality gradual change anomaly based on the feedback verification data.

Images