River ecological health comprehensive evaluation method fusing hydrochemical and biological indexes

Abstract

The invention relates to the technical field of environment monitoring and intelligent management, in particular to a river ecological health comprehensive evaluation method fusing hydrochemistry and biological indexes. The method comprises the following steps: setting a three-point monitoring network in a river, and collecting hydrochemical quick response group parameters and biological activity indication group parameters to form an original monitoring data flow; constructing a single-day and seasonal double-cycle dynamic health baseline by using the original monitoring data stream; extracting a parameter change mode from the double-cycle dynamic health baseline, and setting a health state threshold value; and generating a parameter response mode recognition feature according to the health state threshold value, and forming a dynamic evaluation reference framework. According to the method, the double-cycle dynamic health baseline and the water temperature-season-flow three-dimensional matrix are constructed, so that the evaluation system can automatically adapt to the natural rhythm and real-time hydrological conditions of a river, and the accuracy of abnormal event recognition is improved.

Description

Technical Field

[0001] This invention relates to the field of environmental monitoring and intelligent management technology, and in particular to a comprehensive evaluation method for river ecological health that integrates water chemistry and biological indices. Background Technology

[0002] Traditional river assessment systems still employ static, single-dimensional evaluation standards, which fail to adapt to the dynamic characteristics of river ecosystems. For example, the same dissolved oxygen level may represent a polluted state during the high temperatures of summer, but a healthy state during winter; the diurnal variation pattern is ignored, leading to a large number of false positives or false negatives in the assessment results; and rivers in different basins and under different hydrological conditions are evaluated using the same standards, which lacks scientific rigor.

[0003] On the other hand, traditional water environment monitoring lacks the ability to quickly identify pollution types and accurately locate pollution sources after anomalies are detected. Existing technologies can usually only confirm water quality anomalies, but with thousands of pollutants, it is difficult to determine the specific source of pollution based on limited indicators alone; the lack of multi-point linkage analysis between upstream and downstream makes it impossible to trace the pollution path; and reliance on manual sampling and laboratory analysis prolongs the response cycle to several hours or even days, missing the best time for treatment.

[0004] Most critically, there is a serious disconnect between environmental monitoring and environmental management. The massive amounts of data obtained by monitoring departments are difficult to translate into decision-making basis for management departments; there is a lack of standardized connection mechanisms between data analysis and management actions; environmental law enforcement personnel rely heavily on experience-based judgment, and the allocation of emergency response resources lacks scientific guidance; the effectiveness evaluation and system optimization after environmental incidents lack a closed-loop feedback mechanism, leading to the recurrence of similar problems.

[0005] In summary, the core problem that urgently needs to be solved in existing technologies is: how to establish a dynamic and adaptive evaluation system to achieve rapid pollution identification and accurate source tracing, and seamlessly transform monitoring data into a closed-loop system for management decision-making. Summary of the Invention

[0006] Therefore, it is necessary to provide a comprehensive evaluation method for river ecological health that integrates water chemistry and biological indices to solve at least one of the aforementioned technical problems.

[0007] To achieve the above objectives, a comprehensive evaluation method for river ecological health integrating water chemistry and biological indices is proposed, comprising the following steps: Step S1: Set up a three-point monitoring network in the river, collect water chemistry rapid response parameters and biological activity indicator parameters to form the raw monitoring data stream; Step S2: Construct a dual-cycle dynamic health baseline for daily and seasonal data using the raw monitoring data stream; extract parameter change patterns from the dual-cycle dynamic health baseline and set health status thresholds; generate parameter response pattern recognition features based on the health status thresholds to form a dynamic evaluation benchmark framework. Step S3: Identify abnormal river events based on the dynamic evaluation benchmark framework and construct a pollution feature database; perform feature matching between the abnormal river event data and the pollution feature database to determine the pollution type and estimate the location of the pollution source; generate a management decision support table; Step S4: Activate the graded response mechanism according to the decision support table and record the response effectiveness data. Generate an effectiveness report based on the response effectiveness data and update the pollution feature database and decision-making scheme.

[0008] The beneficial effects of this invention are as follows: By constructing a dual-cycle dynamic health baseline and a three-dimensional matrix of water temperature, season, and flow, the evaluation system can automatically adapt to the natural rhythm of the river and real-time hydrological conditions, thereby greatly improving the accuracy of abnormal event identification and effectively avoiding misjudgments and omissions under traditional static standards. Based on this, by establishing a pollution feature database containing characteristics of three stages—initial response period, rapid change period, and stable period—and combining it with spatiotemporal data from a three-point monitoring network for feature matching, rapid and accurate diagnosis of pollution types and efficient upstream location of pollution sources are achieved, completely changing the previous situation of blind emergency response and difficult investigation and evidence collection. More importantly, this method automatically transforms the diagnostic results into a management decision support table containing graded response parameters, directly driving scientific and efficient emergency response actions, achieving seamless integration of monitoring data to management decisions and optimized allocation of regulatory resources. Simultaneously, the closed-loop feedback mechanism based on response performance data ensures continuous self-optimization of the pollution feature database and decision-making schemes, enabling the entire management system to possess self-learning and evolutionary capabilities. Attached Figure Description

[0009] Figure 1 A schematic diagram of the steps in a comprehensive evaluation method for river ecological health that integrates water chemistry and biological indices; Figure 2 This is a schematic diagram of the three-point monitoring network in this invention; Figure 3 This is a schematic diagram of the construction of the dynamic health baseline in this invention. Detailed Implementation

[0010] The technical method of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0011] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0012] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0013] To achieve the above objectives, please refer to Figures 1 to 3 This invention provides a comprehensive evaluation method for river ecological health that integrates water chemistry and biological indices, comprising the following steps: Step S1: Set up a three-point monitoring network in the river, collect water chemistry rapid response parameters and biological activity indicator parameters to form the raw monitoring data stream; In one embodiment, a three-point monitoring network is deployed in the river according to the source-middle-sink point to collect water chemical and biological activity parameters synchronously at a unified frequency; the collected data is transmitted in real time through the Internet of Things and undergoes a data cleaning process of outlier filtering, smoothing and short-term missing data imputation to finally form a high-quality raw monitoring data stream.

[0014] Step S2: Construct a dual-cycle dynamic health baseline for daily and seasonal data using the raw monitoring data stream; extract parameter change patterns from the dual-cycle dynamic health baseline and set health status thresholds; generate parameter response pattern recognition features based on the health status thresholds to form a dynamic evaluation benchmark framework. In one embodiment, a dual-cycle dynamic health baseline is constructed using historical health data. By calculating the standard daily variation curves of each parameter and the seasonal adjustment factor, a dynamic baseline value and its ±2 standard deviation constitute the health fluctuation range for any given time, thereby setting a health status threshold. Simultaneously, a three-dimensional data matrix of water temperature, season, and flow is established. Based on real-time hydrological and meteorological conditions, weight adjustment coefficients are extracted from the matrix to dynamically adjust the evaluation weights of each parameter, forming an adaptive dynamic evaluation benchmark framework.

[0015] Step S3: Identify abnormal river events based on the dynamic evaluation benchmark framework and construct a pollution feature database; perform feature matching between the abnormal river event data and the pollution feature database to determine the pollution type and estimate the location of the pollution source; generate a management decision support table; In one embodiment, real-time monitoring data is compared with a dynamic evaluation benchmark framework to identify abnormal river events, and multi-dimensional time-series data of abnormal periods are extracted to form event slices. By performing similarity matching between the multi-dimensional feature vectors extracted from the event slices and a pre-constructed pollution feature library that quantifies multiple pollution response modes, the pollution type is determined. Combining the time difference of multi-point monitoring and river hydrological information, the location of the pollution source is further traced back, and finally a management decision support table containing diagnostic results and location information is automatically generated.

[0016] Step S4: Activate the graded response mechanism according to the decision support table and record the response effectiveness data. Generate an effectiveness report based on the response effectiveness data and update the pollution characteristic database and decision-making scheme. In one embodiment, based on the pollution type and severity determined in the management decision support table, graded response parameters for command allocation are automatically generated to activate the corresponding emergency response mechanism. During the response process, performance indicators such as response start time and diagnostic accuracy are automatically recorded. After the incident is handled, the actual pollution type confirmed by on-site investigation is used to compare, correct, or add to the pollution feature database to achieve closed-loop optimization of the evaluation and decision-making model.

[0017] Preferably, step S1 includes the following steps: Step S11: Set up monitoring points in the river in the order of source-middle-sink, with a spacing of 1-2km for short rivers, 3-5km for medium rivers, and 8-10km for long rivers; Step S12: Collect river ecological parameters, transmit the river ecological parameters to the central platform through the Internet of Things module and mark the timestamp and site ID; Step S13: Perform regular filtering and smoothing on outliers in river ecological parameters, and use linear interpolation to fill in short-term missing values ​​in river ecological parameters; Step S14: Simultaneously collect water chemical parameters and biological activity parameters, and establish spatiotemporal pairing relationships.

[0018] In one embodiment, the operation of setting up a three-point monitoring network is as follows: A geographic information system (GIS) is used to digitally analyze the target river, and combined with hydrological data and historical pollution source distribution data, the specific locations of monitoring sections are determined. The source point, i.e., the upstream control point, is set in an area upstream of the target river segment that is not significantly affected by human activities, used to obtain background water quality data of the river. The midpoint, i.e., the midpoint of the target river segment, is set in the center or downstream of the potential pollution source's influence area, used to capture the core response of pollution events. The sink point, i.e., the downstream response point, is set at a sufficient distance downstream of the target river segment, used to assess the diffusion and self-purification processes of pollutants. The specific spacing between each monitoring point is determined according to the river's classification: for short rivers less than 50 km in length, the spacing is set to 1-2 km; for medium-length rivers between 50-150 km in length, the spacing is set to 3-5 km; and for long rivers greater than 150 km in length, the spacing is set to 8-10 km.

[0019] In one embodiment, the data acquisition and transmission operation is as follows: An online monitoring device integrating multi-parameter sensors is deployed at each monitoring point. The device's built-in IoT module includes a microcontroller, a real-time clock module, and a 4G / 5G communication module. The microcontroller polls and reads sensor values ​​for the rapid response group parameters of water chemistry and the biological activity indicator group parameters at a preset frequency of once every 15 minutes. After each full parameter acquisition, the microcontroller obtains the current precise time from the real-time clock module and, combined with the pre-configured unique site identification ID in the device, packages all acquired parameter values, timestamps, and site IDs into a data frame. Subsequently, the data frame is sent to the central data platform via the 4G / 5G communication module using the TCP / IP protocol.

[0020] In one embodiment, the outlier filtering and smoothing operation is as follows: After receiving the data frame, the central data platform initiates a data cleaning procedure for the hydrochemical rapid response group parameters and the biological activity indicator group parameters. First, outlier filtering is performed. This procedure maintains a sliding window based on the first 24 data points for each parameter and calculates the standard deviation of the data within the window. and mean If the value of the newly received data point exceeds... If the value falls within the specified range, the data point is marked as an isolated outlier and discarded. Next, data smoothing is performed using a 5-point moving average method, replacing the current data point's value with the arithmetic mean of itself and the two data points before and after it (a total of 5 points), thereby eliminating short-term random noise in the sensor signal.

[0021] In one implementation of this embodiment, the operation of establishing spatiotemporal pairing is as follows: In step S12, the microcontroller sequentially reads all water chemical and biological activity parameter sensors at extremely short time intervals (milliseconds) within a single acquisition cycle, and associates this set of data with a unified timestamp. On the central data platform side, the site ID and timestamp in each received data frame are combined as a composite primary key for the database. Through this composite primary key, it is ensured that all parameter values ​​collected at the same time and location are strictly paired and stored, forming the basic data units required for subsequent multi-parameter joint analysis and pattern recognition.

[0022] Preferably, step S2 includes the following steps: Step S21: Select health status historical data for more than one year, divide the health status historical data into time points, calculate the multi-year average and standard deviation of each point, and generate a standard daily variation curve; Step S22: Calculate the offset of historical health status data relative to the annual average, and generate seasonal adjustment parameters; Step S23: Integrate the standard daily variation curve with the seasonal adjustment parameters to construct a dynamic baseline and its ±2 standard deviation range, forming a river health sequence; Step S24: Analyze the watershed characteristics of the river health sequence; establish a three-dimensional matrix of water temperature-season-flow based on the watershed characteristics, and dynamically adjust the parameter weight coefficients using the three-dimensional matrix of water temperature-season-flow.

[0023] In one embodiment, the operation of generating a standard diurnal variation curve is as follows: From the historical raw monitoring data stream, select health status historical data for at least 12 consecutive months that have been confirmed as pollution-free events by expert review or historical event records. Divide each 24-hour day into 96 fixed time points at 15-minute intervals. For each time point (e.g., the 48th time point represents 12:00 noon), extract the parameter value for that time point from all historical health status data, and calculate the multi-year arithmetic mean and standard deviation of these values. Connect the multi-year averages of these 96 time points in chronological order to form a smooth standard diurnal variation curve representing the typical diurnal variation pattern of the parameter. Simultaneously, store the standard deviation values ​​of the 96 time points as the basis for subsequently constructing the health fluctuation range.

[0024] In one embodiment, the operation of generating the seasonal adjustment parameter is as follows: the historical health status data is grouped by month (January to December). The monthly average of all data for each month is calculated, and the annual average of the entire historical health status data is also calculated simultaneously. For each month, the seasonal adjustment parameter is calculated by subtracting the annual average from the monthly average. This results in an array containing 12 offset values, namely the "seasonal adjustment parameter," which quantifies the degree to which each month is systematically higher or lower than the annual average.

[0025] In one embodiment, the operation of constructing a dynamic baseline and its ±2 standard deviation range is as follows: For any 15-minute time point on any future day, the dynamic baseline value is generated as follows: First, the multi-year average value corresponding to that time point is found from the standard daily variation curve. Then, based on the month to which the date belongs, the corresponding monthly offset is found from the seasonal adjustment parameters. The multi-year average value and the monthly offset are added together, and the result is the "dynamic baseline value" at that moment. Next, the standard deviation σ corresponding to that time point is obtained from the standard deviation value stored in step S21. By adding or subtracting twice the standard deviation from the dynamic baseline value (i.e., dynamic baseline value ±2σ), a dynamic "health fluctuation range" is constructed. The dynamic baseline values ​​of all time points are arranged in chronological order to form the river health sequence.

[0026] In one embodiment, the operation of establishing a three-dimensional matrix of water temperature, season, and flow rate is as follows: First, water temperature and flow rate data of the watershed are collected from historical data. The water temperature data is divided into three levels based on preset thresholds: high-temperature period (e.g., >25°C), medium-temperature period (e.g., 15-25°C), and low-temperature period (e.g., <15°C). The flow rate data is divided into three levels based on historical flow percentiles: high-water period (e.g., flow rate >75th percentile), normal-water period (e.g., 25%-75th percentile), and low-water period (e.g., <25th percentile). Combining the 12 natural months, a three-dimensional data matrix is ​​constructed with season (month), water temperature level, and flow rate level as the three axes.

[0027] It should be noted that the operation of dynamically adjusting the parameter weight coefficients using the aforementioned three-dimensional data matrix of water temperature, season, and flow rate is as follows: a set of weight adjustment coefficients is preset for each node in the three-dimensional data matrix (representing a specific combination of water temperature, season, and flow rate). These coefficients are determined based on expert knowledge and historical data analysis. For example, in the node representing "summer-high temperature period-dry season," the weight adjustment coefficient for dissolved oxygen is set to 1.15 (indicating a 15% increase in weight) to enhance sensitivity to the risk of high temperature and hypoxia; in the node representing "rainy season-any temperature period-high water period," the weight adjustment coefficient for turbidity is set to 0.90 (indicating a 10% decrease in weight) to reduce non-polluting high turbidity interference caused by storm runoff. During real-time evaluation, based on the currently monitored real-time water temperature and flow rate data, a unique corresponding node is located in the three-dimensional data matrix, and the weight adjustment coefficient of that node is extracted. This is used to dynamically weight the basic weights of each parameter, thereby generating the final real-time weight coefficients used for health status calculation.

[0028] Preferably, step S23 includes: Extract the baseline value for each time point in the standard daily variation curve; The adjustment factor for each month is calculated based on historical health status data, and the baseline value is seasonally corrected. Sliding window smoothing is applied to data changes between adjacent time points; Calculate the standard deviation at each time point, and construct the range of health fluctuations using the baseline value ± 2 times the standard deviation; The baseline values ​​are updated quarterly, and the dynamic baseline parameters are adjusted based on newly added monitoring data.

[0029] In one embodiment, the operation of extracting the benchmark value is as follows: directly read the multi-year arithmetic mean stored at the 15-minute time point corresponding to the current moment from the generated standard diurnal variation curve. This multi-year arithmetic mean is defined as the unadjusted benchmark value at that moment.

[0030] In one embodiment, the seasonal correction of the baseline value is performed as follows: First, by calculating the difference between the monthly average and the annual average, a seasonal adjustment parameter array containing 12 offsets is obtained. For any given time point to be calculated, the corresponding monthly adjustment coefficient (i.e., offset) is retrieved from this array based on the month it falls in. The baseline value extracted in the previous step is algebraically summed with this monthly adjustment coefficient, and the result is the seasonally corrected dynamic baseline value.

[0031] Using the current weighting adjustment coefficients to calculate the health index, a seasonal adaptability score is obtained. In one embodiment, the operation of constructing the health fluctuation range is as follows: First, the standard deviation σ corresponding to the current moment is read from the standard deviation values ​​of the 96 time points stored together when constructing the standard daily variation curve. Then, the dynamic baseline value after seasonal correction and smoothing is used as the center line. The upper limit of the health fluctuation range is obtained by adding twice the standard deviation to the dynamic baseline value, and the lower limit of the health fluctuation range is obtained by subtracting twice the standard deviation from the dynamic baseline value. The interval formed by these two upper and lower limits is the health fluctuation range at that moment.

[0032] In one embodiment, the quarterly update operation involves pre-setting a baseline update procedure that is automatically triggered at the end of each quarter (e.g., March 31, June 30, etc.). This procedure merges all newly added monitoring data confirmed as healthy within the past quarter into the existing historical health database. Subsequently, the complete calculation process of constructing the standard diurnal variation curve (step S21) and generating seasonal adjustment parameters (step S22) is re-executed, thereby generating a new version of the standard diurnal variation curve (containing new baseline values) and new seasonal adjustment parameters that include the latest data information. This new set of dynamic baseline parameters will be used for subsequent river health assessments starting from the first day of the next quarter.

[0033] Preferably, step S24 includes: Historical water temperature and flow data of the watershed were collected, and a three-dimensional data matrix was constructed by combining seasonal adjustment parameters. The horizontal axis represents the season, the vertical axis represents water temperature, and the Z-axis represents flow. Assign a weight adjustment coefficient to each node in the three-dimensional data matrix, where the rule for assigning the weight adjustment coefficient is to increase the weight of dissolved oxygen during high-temperature periods and decrease the weight of turbidity during rainfall periods. Based on real-time monitored water temperature and flow data, extract the current weight adjustment coefficients from the three-dimensional data matrix; The health index is calculated using the current weighting adjustment coefficients to obtain a seasonal adaptability score.

[0034] In one embodiment, the operation of constructing a three-dimensional data matrix is ​​as follows: First, daily water temperature and flow data of the target watershed over the past few years are extracted from the historical monitoring database. The water temperature data is divided into three discrete levels based on preset thresholds: high temperature (e.g., >25℃), moderate temperature (e.g., 15-25℃), and low temperature (e.g., <15℃). The flow data is divided into three discrete levels based on the percentiles of its historical data: abundant water (e.g., >75th percentile), normal water (e.g., 25%-75th percentile), and low water (e.g., <25th percentile). The year is divided into four seasons: spring (March-May), summer (June-August), autumn (September-November), and winter (December-February). Using these three dimensions—season (4 levels), water temperature level (3 levels), and flow level (3 levels)—a discrete three-dimensional data matrix containing 4×3×3=36 nodes is constructed.

[0035] In one embodiment, the operation of assigning weight adjustment coefficients is as follows: based on expert knowledge and historical data analysis, a set of weight adjustment coefficients containing all evaluation parameters is preset for each node in the three-dimensional data matrix. For example, for a node representing the "summer-high temperature-dry season" combination, the weight adjustment coefficient for the dissolved oxygen parameter is set to 1.15 to increase its weight in the evaluation; for a node representing the "summer-medium temperature-abundant water season (usually corresponding to the rainfall period)" combination, the weight adjustment coefficient for the turbidity parameter is set to 0.90 to reduce its weight. All other parameters under these specific conditions have a weight adjustment coefficient of 1.00, indicating no adjustment. This process assigns a complete set of weight adjustment coefficients to all 36 nodes in the matrix.

[0036] In one embodiment, the operation of extracting the current weight adjustment coefficients is as follows: When performing a real-time health assessment, the real-time water temperature and real-time flow rate values ​​are first obtained. Based on the same threshold classification rules described above, the real-time water temperature value is mapped to one of three levels: high temperature, medium temperature, or low temperature; the real-time flow rate value is mapped to one of three levels: abundant water, normal water, or low water. Simultaneously, the season is determined based on the current date. Using these three determined dimensional coordinates (season, water temperature level, and flow rate level), a node is uniquely located in the three-dimensional data matrix, and the preset set of weight adjustment coefficients is extracted from that node.

[0037] In one implementation of this embodiment, the operation of calculating the health index using the current weight adjustment coefficient is as follows: First, the real-time monitoring value of each evaluation parameter is normalized to obtain the normalized parameter value. Then, the base weights of each parameter are... The corresponding weight adjustment coefficient extracted in the previous step Multiply to obtain the dynamically adjusted real-time weights. Finally, by weighting and summing the normalized values ​​of all parameters with their corresponding real-time weights, the final health index HI is calculated, which is the seasonal adaptability score.

[0038] Preferably, step S2 further includes the following steps: Anomaly response analysis is performed on the parameter change rate and duration in the raw monitoring data stream to form anomaly response events, which are then classified into abrupt and gradual types. Construct a time-series correlation matrix among parameters in the raw monitoring data stream to identify the synergistic variation characteristics of specific pollution; Extract multi-parameter change features of abnormal response events and combine them with collaborative change features to form a monitoring feature vector; The monitoring feature vectors are stored in the time series feature library as reference data for pollution feature matching; Based on the historical accumulated data of the time series feature library, a parameter covariance table is established.

[0039] In one embodiment, the anomaly response analysis operation is as follows: when the real-time value of any parameter deviates from its dynamic baseline by more than a preset threshold, an anomaly response analysis procedure is triggered. This procedure first calculates the rate of change of the parameter at each subsequent time point. The rate of change is calculated by dividing the difference between the current value and the value at the previous time point by the time interval. If the rate of change exceeds a preset abrupt change threshold within 5 minutes (e.g., a 30% change in parameter value), the event is marked as a "mutant" response event. If the rate of change does not reach the abrupt change threshold, but the cumulative change in the parameter value within 30 minutes exceeds a preset gradual change threshold (e.g., a 50% change in parameter value), the event is marked as a "gradual" response event. The duration of the event is recorded as the time from the first deviation from the baseline to the return to the baseline within the fluctuation range.

[0040] In one embodiment, the operation of constructing the time-series correlation matrix is ​​as follows: Extract a fixed-length time window (e.g., one hour before and after the identified anomalous response event) of raw monitoring data stream. For all evaluation parameters within this time window, calculate the Pearson correlation coefficient between the time series of any two parameters (e.g., dissolved oxygen and bacterial ATP activity). Organize the correlation coefficient calculation results of all parameter pairs into a symmetric N×N matrix, where N is the total number of evaluation parameters. This matrix is ​​the time-series correlation matrix between parameters. Coefficients with absolute values ​​close to 1 in the matrix (e.g., >0.8 or <-0.8) indicate that the two parameters exhibit strong co-variation characteristics during the anomalous event, such as a strong positive or negative correlation.

[0041] In one embodiment, the operation of forming the monitoring feature vector is as follows: for each identified abnormal response event, a set of numerical values ​​quantifying its characteristics is automatically extracted. This set of values ​​includes: the type of the event (abrupt or gradual), the maximum rate of change of each parameter, the duration of change of each parameter, and all strongly co-changing parameter pairs and their correlation coefficients extracted from the time-series correlation matrix. These values ​​are arranged in a predefined order to form a fixed-dimensional vector, which is the "monitoring feature vector" of this abnormal response event.

[0042] In one implementation of this embodiment, the operation of storing data in the time-series feature library involves maintaining a time-series feature database to store the monitoring feature vectors of all historical abnormal response events. Whenever a new monitoring feature vector is generated, it is appended with metadata such as a unique event ID, occurrence time, and the monitoring station to which it belongs, and then stored as a new record in the database. This database is the "time-series feature library," which provides the basic data source for subsequent construction and optimization of pollution features through machine learning algorithms, or for direct case comparison.

[0043] In one embodiment, the operation of establishing the parameter covariance table involves periodically (e.g., monthly) performing statistical analysis on all monitoring feature vectors in the time-series feature database. By analyzing the time-series correlation matrix data of a large number of events in the database, the frequency of strong covariance changes (e.g., absolute correlation coefficient > 0.8) between different parameter pairs is statistically determined. The most frequent parameter pairs and their most common correlation directions (positive or negative correlation) are recorded to form a "parameter covariance table." This table reveals which parameters are most likely to change in tandem with disturbances in the local river environment, providing a basis for quickly identifying potential associated indicators of pollution events.

[0044] Preferably, step S3 includes the following steps: Step S31: Obtain and construct a multi-dimensional pollution feature database through historical pollution event analysis and river section simulation; Step S32: Calculate the multi-gradient concentration of each type of pollutant using a multi-dimensional pollution feature library to form data on the relationship between concentration and parameter response; Step S33: Based on the concentration and parameter response data, the pollution response process is divided into the initial response period, the rapid change period, and the stable period, and characteristic parameters of each stage are extracted; Step S34: Encode the pollutant features for each pollutant and convert the response features of each pollutant into a 64-bit digital feature code; Step S35: Verify the consistency of digital signature codes under different water temperatures and flow rates, and supplement and improve the multidimensional pollution signature database.

[0045] In one embodiment, the process of constructing a multidimensional pollution feature database is as follows: First, historically confirmed pollution event records are collected and organized, with each record containing the type of pollutant, emission concentration, and full-parameter monitoring data during the event. Second, simulated experiments are conducted in controlled laboratory tanks or isolated experimental river sections, injecting known types and concentrations of pollutants (such as high-concentration organic wastewater, copper-containing heavy metal solutions, specific pesticides, etc.) and recording the full-parameter response process using the same monitoring equipment as in the field. These data from real events and simulated experiments are then aggregated to form a raw database for constructing pollution features. This database is the prototype of the "multidimensional pollution feature database."

[0046] In one embodiment, the operation of forming concentration-parameter response data involves: filtering all relevant event data from the original database of the multidimensional pollution feature library for a specific type of pollutant (e.g., organic pollution). These events are then divided into multiple gradients based on pollutant concentration, such as low, medium, and high concentration levels. For each concentration level, the average shape of the response curves for each monitoring parameter (e.g., dissolved oxygen, bacterial ATP activity) in all corresponding events is calculated. Through this process, the correspondence data of the response amplitude and response speed of each monitoring parameter caused by each type of pollutant at different pollution intensities is established.

[0047] In one embodiment, the operation of extracting characteristic parameters for each stage is as follows: Each average response curve generated in the previous step is divided into stages. The starting point of the "initial response period" is defined as the time point when the parameter value first and continuously deviates from its dynamic baseline. The period when the parameter value's rate of change reaches its maximum is defined as the "rapid change period." The period when the parameter value's rate of change approaches zero and the value enters a new stable or slow recovery plateau is defined as the "stabilization period." Then, a set of key characteristic parameters is extracted from these three stages, including: the lag time of the initial response, the peak rate of change and duration of the rapid change period, and the final deviation of the parameter value from the baseline in the stabilization period.

[0048] In one embodiment, the conversion to a 64-bit digital feature code involves constructing a 64-bit binary encoding structure for each pollution type. This structure is divided into multiple fields, each storing a specific response characteristic. For example, bits 1-8 encode the dominant response indicator (which parameter responds first); bits 9-16 encode the magnitude of the peak change rate; and bits 17-24 encode the response time-series relationship (e.g., the lag time of DO decrease relative to ATP increase). All stage feature parameters extracted in step S33 are converted into binary numbers according to a preset quantization rule and filled into the corresponding fields of this 64-bit structure. The resulting 64-bit binary code is the "digital feature code" for that pollution type at a specific concentration level.

[0049] In one embodiment, the verification and supplementation operations are as follows: Pollution events of the same type and concentration level occurring under different water temperatures and flow rates are selected from the original database of the multidimensional pollution feature library. The response characteristics of these events are extracted and their digital feature codes are generated. These digital feature codes generated under different environmental conditions are compared, and the consistency and variation patterns of their core fields are analyzed. If certain fields are found to change systematically with water temperature or flow rate, this variation pattern (e.g., shortened response delay at high temperatures) is used as supplementary information and stored together with the core digital feature code of that pollution type in the final multidimensional pollution feature library, thereby enhancing the environmental adaptability of the features.

[0050] Preferably, step S33 includes: The point at which the marked parameters begin to deviate from the baseline is taken as the starting point of the initial response period, and the lag time of the initial response is recorded. The period during which the rate of change of the marked parameter reaches its maximum value is the period of rapid change. The peak value and duration of the rate of change are extracted. The period during which the marked parameters tend to stabilize is called the stabilization period. The deviation of the stabilization period parameter values ​​from the baseline is calculated. Key characteristic parameters of the three stages—initial response period, rapid change period, and steady period—are extracted, including initial response delay, peak change rate, and steady-state value offset. Analyze the temporal relationships between different stages to establish a temporal response characteristic map of pollutants; Construct a response phase ratio model for each pollutant and record the percentage of the total response time for each of the three phases.

[0051] In one embodiment, the operation of recording the lag time of the initial response is as follows: For a contamination response curve, starting from the start time of contamination injection, the parameter value is checked point by point to see if it exceeds the dynamic health fluctuation range at the corresponding time. When the parameter value is outside the fluctuation range for the first time and continuously (e.g., for three consecutive data points), the timestamp of this first deviation point is marked as the starting point of the "initial response period". The time difference between this timestamp and the start time of contamination injection is recorded as the "initial response delay" time of the parameter.

[0052] In one embodiment, the operation of extracting the peak value and duration of the rate of change is as follows: First, the first derivative of the pollution response curve is calculated to obtain a parameter rate of change curve. The maximum absolute value of this rate of change curve is found; this maximum value is the "peak rate of change". Then, a time interval centered on the peak rate of change and with a width equal to 80% of the peak value's height is defined; the time period corresponding to this interval is marked as the "rapid change period". The length of this time interval is recorded as the "duration of the rapid change period".

[0053] In one embodiment, the operation of calculating the deviation of the stable period parameter value from the baseline is as follows: On the rate of change curve, starting from the end of the rapid change period, a search is performed backwards. When the absolute value of the rate of change first consistently falls below a preset stability threshold (e.g., below 5% of the peak rate of change), this time point is marked as the starting point of the "stable period". A fixed-length segment of data (e.g., 1 hour) starting from the starting point of the stable period is extracted, and the average value of this segment is calculated, which is the "stable period parameter value". The difference between this stable period parameter value and the average value of the dynamic baseline value within the same time period is recorded as the "stable value offset".

[0054] In one implementation of this embodiment, the operation of extracting key feature parameters is as follows: the three values ​​of "initial response delay", "peak change rate" and "stable value offset" calculated for each response parameter in the above steps, as well as the "duration of rapid change period", are used as a set of core quantitative indicators describing the dynamic process of the parameter's response and are stored in a structured manner.

[0055] In one embodiment, the operation of establishing a time-series response feature map is as follows: For a pollution event involving multiple parameter responses, the timestamps of the start of the "initial response period," the start of the "rapid change period," and the start of the "stable period" for all parameters are aligned and compared. By calculating the difference in the initial response delays of different parameters (such as bacterial ATP activity and dissolved oxygen), the order and time difference of their responses are determined. This structured information containing the order and time difference of all parameter responses is stored in graphical or tabular form to form the "time-series response feature map" of the pollution event.

[0056] Preferably, step S3 further includes: When a health index in the dynamic evaluation benchmark framework triggers an early warning, it automatically extracts time-series data of abnormal parameters to form an event slice. Analyze the multidimensional variation characteristics of event slices and calculate the similarity between event slices and features in the multidimensional pollution feature library; Similarity is graded into high match, medium match, and low match, with similarity ≥85% being high match, 75-85% being medium match, and <75% being low match. A high match indicates a single contamination type, a medium match indicates multiple contamination types, and a low match indicates an unknown contamination type.

[0057] In one embodiment, the operation constituting an event slice is as follows: when the real-time monitored Health Index (HI) first falls below a preset warning threshold, the timestamp of that moment is immediately marked as an anomaly trigger point. Centered on this anomaly trigger point, time-series data of all hydrochemical rapid response group parameters and biological activity indicator group parameters for two hours before and after (a total of four hours) are automatically extracted from the original monitoring data stream. This multi-dimensional time-series data set, containing the complete process before, during, and after the anomaly, is defined as an "event slice".

[0058] In one embodiment, the similarity analysis and calculation process is as follows: First, the same feature extraction algorithm used when constructing contamination features is applied to the event slice, i.e., the rate of change, duration, response delay, and temporal correlation between parameters are calculated to generate a "multidimensional feature vector to be matched" representing the current real-time abnormal event. Then, a cosine similarity algorithm is used to calculate the multidimensional feature vector to be matched against the digital feature code (or its corresponding pattern vector) of each contamination type stored in the multidimensional contamination feature database. The cosine similarity is calculated as follows: ; in: This is the similarity score, with a value range of [-1, 1]. The closer the value is to 1, the more similar the similarity. It is the multidimensional feature vector to be matched; It is a pattern vector in the pollution feature library; It is a vector and The dot product; and They are vectors and The Euclidean norm.

[0059] This process will generate a list for the current event containing its similarity scores to all known contamination types.

[0060] In one embodiment, the similarity grading operation is as follows: The similarity score list generated in the previous step is processed. First, the highest similarity score in the list is found. Then, the highest score is graded according to a preset threshold rule: if the highest similarity score is greater than or equal to 0.85, the matching result is marked as "high match". If the highest similarity score is between 0.75 (inclusive) and 0.85 (exclusive), it is marked as "medium match". If the highest similarity score is lower than 0.75, it is marked as "low match".

[0061] In one implementation of this embodiment, the operation of determining the contamination type based on the matching level is as follows: If the matching result is "high match," the contamination type corresponding to the highest similarity score is directly determined as the unique diagnostic result for this abnormal event. If the matching result is "medium match," the top three contamination types with the highest similarity scores are selected and output as a list of candidate diagnostic results for this event. If the matching result is "low match," the diagnostic result for this event is marked as "unknown contamination type," and a flag requiring human expert intervention for analysis is triggered.

[0062] Preferably, step S3 further includes: Record the timestamps of anomalies occurring at upstream and downstream sections in the three-point monitoring network and calculate the time difference; Obtain the distance data between cross sections and calculate the pollutant propagation speed by combining the time difference; Collect real-time flow velocity data and establish a library of propagation correction coefficients based on pollutant characteristics; The location of the pollution source is calculated by tracing back to the source using a propagation correction coefficient library. Match the location of pollution sources with the river estuary information in the geographic information system; Generate spatial coordinates and positioning accuracy estimates of the pollution source location, and output a map of the investigation area.

[0063] In one embodiment, the operation of calculating the time difference is as follows: when the upstream monitoring point (source point) and the downstream monitoring point (midpoint or sink point) successively detect anomalies with the same or highly similar characteristics caused by the same pollution event, the timestamps of the anomaly trigger points of these two monitoring points are recorded respectively. The timestamp of the downstream monitoring point is subtracted from the timestamp of the upstream monitoring point, and the result is the time difference taken for the pollutant plume to propagate between the two monitoring sections. ).

[0064] In one embodiment, the operation of calculating the pollutant propagation speed is as follows: precise river channel distance data between two monitoring sections where anomalies have occurred is read from a pre-configured geographic information database. Then, by dividing the river channel distance data by the time difference calculated in the previous step, the apparent propagation velocity of the pollutant plume is obtained. The calculation method is as follows: ; In one embodiment, the operation of establishing the propagation correction coefficient library is as follows: First, by installing equipment such as acoustic Doppler current meters at monitoring points, the average flow velocity data of the river channel is collected in real time. Secondly, based on fluid dynamics principles and historical experimental data, a "propagation correction coefficient library" was established for different types of pollutants. This library stores the ratio of the propagation velocity to the water flow velocity (i.e., the propagation correction coefficient) of different pollutants (such as dissolved pollutants and suspended pollutants) under different flow velocities and riverbed roughness conditions. For example, dissolved pollutants The value is close to 1, while the suspended pollutants that are easy to settle... The value is less than 1.

[0065] In one implementation of this embodiment, the upstream positioning operation is as follows: First, based on the identified pollution type, the corresponding propagation correction coefficient is retrieved from the propagation correction coefficient library. Then, by using the average flow rate data collected in real time... Multiply by the correction factor Calculate the actual propagation speed of the pollutants ( ).like Compared with the previously calculated apparent propagation speed If there are significant differences, then it will be based on For accuracy, the distance to the pollution source is estimated by starting from the monitoring point where the anomaly occurred (usually the upstream anomaly point in the three-point network) and moving upstream along the river. Calculated in the following way: ; in, It is the time from the start of the pollution event (which can be inferred from the shape of the response curve) to the time when the anomaly is detected by the monitoring point.

[0066] In one embodiment, the matching operation is as follows: the estimated distance calculated in the previous step is used as the basis for the matching operation. On the river vector map of the Geographic Information System (GIS), the monitoring point where the anomaly occurred is calibrated upstream to determine the estimated geographic coordinates of the pollution source emission point. Then, within a preset radius (e.g., 500 meters) around these geographic coordinates, the location information of all known sewage outlets, storm drains, or key risk enterprises discharging into the river is searched.

[0067] In one embodiment, the operation of generating spatial coordinates and outputting the map is as follows: all potential sewage outlets matched in the previous step are designated as high-priority investigation targets. The geographic coordinates, names, and related information of these targets are listed. Simultaneously, based on the uncertainty of propagation speed calculation and the range of GIS matching, an estimate of positioning accuracy is given (e.g., "located within 3.5 ± 0.5 km upstream"). Finally, an "investigation area map" containing the river section, the location of abnormal monitoring points, and the markers of all high-priority investigation targets is generated and output as part of the management decision support table.

[0068] Please see Figure 2 The diagram shows a three-point monitoring network in this invention. The diagram clearly shows the three-point monitoring network deployed along a horizontal river. By setting up an upstream control point, the midpoint of the target river section, and a downstream response point, combined with the arrows indicating the direction of water flow, the spatial layout of water quality monitoring is presented intuitively.

[0069] Please see Figure 3 This is a schematic diagram of the construction of the dynamic health baseline in this invention. This diagram uses a two-dimensional coordinate system, with the standard daily variation curve as a reference, and overlays the summer dynamic baseline and the dashed line of the health fluctuation range, clearly reflecting the dynamic changes of parameter values ​​and the definition of health intervals at different time scales.

[0070] Of particular importance, step S4 includes: A Level III emergency response mechanism will be activated based on the degree of decline in the health index. Increase monitoring frequency and activate automatic sampling for mild anomalies; In cases of moderate abnormalities, specialized evidence collection equipment and personnel should be deployed. In cases of severe abnormalities, a multi-departmental coordination mechanism and specialized testing equipment are activated; Record performance metrics such as time from anomaly detection to response initiation and recognition accuracy. Continuously monitor the water quality recovery process until the indicators return to the preset river health threshold and remain stable for 24 hours.

[0071] In one embodiment, the operation of activating the three-level emergency response mechanism is as follows: First, the severity of the event is calculated based on the maximum deviation of parameter values ​​from their dynamic baseline and the duration during the abnormal event. Then, according to a preset severity threshold, the event is automatically divided into three response levels: if the deviation is less than 20% and the duration is less than 1 hour, it is defined as "mild anomaly"; if the deviation is between 20% and 40% or the duration is between 1 and 3 hours, it is defined as "moderate anomaly"; if the deviation is greater than 40% or specific high-risk pollutant characteristics are detected, it is defined as "severe anomaly". The corresponding response procedure is automatically activated according to this level.

[0072] In one implementation of this embodiment, the response to a "minor anomaly" is as follows: A parameter modification command is automatically sent to the monitoring device that triggered the anomaly, changing the monitoring data acquisition frequency parameter of that device from 15 minutes to 3 minutes. Simultaneously, a sampling start command is sent to the automatic sampler integrated into the monitoring point. This command includes the sampling bottle number and the current timestamp, to retain the water sample for subsequent laboratory verification.

[0073] In another embodiment, the response operation for "moderate anomaly" is as follows: when generating the management decision support table, a set of resource allocation parameters is automatically generated. For example, the number of law enforcement personnel is generated as 3; and based on the identified pollution type, an emergency evidence collection equipment type parameter is generated. For example, when the pollution type is organic pollution, the equipment type parameter is specified as "portable COD rapid tester", and when it is heavy metal pollution, it is specified as "portable heavy metal detector".

[0074] In another embodiment, the response to a “severe anomaly” is as follows: automatically generate a top-level emergency response identifier, and send a multi-departmental linkage request instruction containing the event ID, the diagnosed pollution type, the estimated location of the pollution source, and the severity assessment to the preset higher-level emergency management information system through a standard data interface.

[0075] It should be noted that the operation of recording performance indicators is as follows: automatically record the timestamp of the first detection of an abnormal event. and the timestamps of when the management decision support form was generated and sent. The response startup time is... and The difference lies in the type of pollution. After the incident is resolved, the management personnel will enter the actual pollution type confirmed through on-site verification into the system. This actual pollution type will be compared with the pollution type diagnosed by the system. If they match, the identification is considered accurate, and the accuracy rate will be calculated accordingly.

[0076] In one embodiment, the operation of monitoring the water quality recovery process is as follows: after the emergency response is initiated, a continuous recovery status monitoring mode is entered. In this mode, the real-time values ​​of all parameters are continuously compared with the health fluctuation range of their corresponding dynamic health baseline. When all parameter values ​​return to their respective health fluctuation ranges for the first time, a 24-hour stabilization period timer is started. If no parameter exceeds its health fluctuation range again within these 24 hours, the status of the abnormal event is automatically marked as "closed" after the timer expires.

[0077] Therefore, the embodiments should be considered as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the application are intended to be included within the invention.

[0078] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A comprehensive evaluation method for river ecological health integrating water chemistry and biological indices, characterized in that, The method includes the following steps: Step S1: Set up a three-point monitoring network in the river, collect water chemistry rapid response parameters and biological activity indicator parameters to form the raw monitoring data stream; Step S2: Construct a dual-cycle dynamic health baseline for daily and seasonal data using the raw monitoring data stream; extract parameter change patterns from the dual-cycle dynamic health baseline and set health status thresholds; generate parameter response pattern recognition features based on the health status thresholds to form a dynamic evaluation benchmark framework. Step S3: Identify abnormal river events based on the dynamic evaluation benchmark framework and construct a pollution feature database; perform feature matching between the abnormal river event data and the pollution feature database to determine the pollution type and estimate the location of the pollution source; generate a management decision support table; Step S4: Activate the graded response mechanism according to the decision support table and record the response effectiveness data. Generate an effectiveness report based on the response effectiveness data and update the pollution feature database and decision-making scheme.

2. The comprehensive evaluation method for river ecological health integrating hydrochemical and biological indices according to claim 1, characterized in that, Step S1 includes the following steps: Step S11: Set up monitoring points in the river in the order of source-middle-sink, with a spacing of 1-2km for short rivers, 3-5km for medium rivers, and 8-10km for long rivers; Step S12: Collect river ecological parameters, transmit the river ecological parameters to the central platform through the Internet of Things module and mark the timestamp and site ID; Step S13: Perform regular filtering and smoothing on outliers in river ecological parameters, and use linear interpolation to fill in short-term missing values ​​in river ecological parameters; Step S14: Simultaneously collect water chemical parameters and biological activity parameters, and establish spatiotemporal pairing relationships.

3. The comprehensive evaluation method for river ecological health integrating hydrochemical and biological indices according to claim 1, characterized in that, Step S2 includes the following steps: Step S21: Select health status historical data for more than one year, divide the health status historical data into time points, calculate the multi-year average and standard deviation of each point, and generate a standard daily variation curve; Step S22: Calculate the offset of historical health status data relative to the annual average, and generate seasonal adjustment parameters; Step S23: Integrate the standard daily variation curve with the seasonal adjustment parameters to construct a dynamic baseline and its ±2 standard deviation range, forming a river health sequence; Step S24: Analyze the watershed characteristics of the river health sequence; establish a three-dimensional matrix of water temperature-season-flow based on the watershed characteristics, and dynamically adjust the parameter weight coefficients using the three-dimensional matrix of water temperature-season-flow.

4. The comprehensive evaluation method for river ecological health integrating hydrochemical and biological indices according to claim 3, characterized in that, Step S23 includes: Extract the baseline value for each time point in the standard daily variation curve; The adjustment factor for each month is calculated based on historical health status data, and the baseline value is seasonally corrected. Sliding window smoothing is applied to data changes between adjacent time points; Calculate the standard deviation at each time point, and construct the range of health fluctuations using the baseline value ± 2 times the standard deviation; The baseline values ​​are updated quarterly, and the dynamic baseline parameters are adjusted based on newly added monitoring data.

5. The comprehensive evaluation method for river ecological health integrating hydrochemical and biological indices according to claim 3, characterized in that, Step S24 includes: Historical water temperature and flow data of the watershed were collected, and a three-dimensional data matrix was constructed by combining seasonal adjustment parameters. The horizontal axis represents the season, the vertical axis represents water temperature, and the Z-axis represents flow. Assign a weight adjustment coefficient to each node in the three-dimensional data matrix, where the rule for assigning the weight adjustment coefficient is to increase the weight of dissolved oxygen during high-temperature periods and decrease the weight of turbidity during rainfall periods. Based on real-time monitored water temperature and flow data, extract the current weight adjustment coefficients from the three-dimensional data matrix; The health index is calculated using the current weighting adjustment coefficients to obtain a seasonal adaptability score.

6. The comprehensive evaluation method for river ecological health integrating hydrochemical and biological indices according to claim 1, characterized in that, Step S2 also includes the following steps: Anomaly response analysis is performed on the parameter change rate and duration in the raw monitoring data stream to form anomaly response events, which are then classified into abrupt and gradual types. Construct a time-series correlation matrix among parameters in the raw monitoring data stream to identify the synergistic variation characteristics of specific pollution; Extract multi-parameter change features of abnormal response events and combine them with collaborative change features to form a monitoring feature vector; The monitoring feature vectors are stored in the time series feature library as reference data for pollution feature matching; Based on the historical accumulated data of the time series feature library, a parameter covariance table is established.

7. The comprehensive evaluation method for river ecological health integrating water chemistry and biological indices according to claim 1, characterized in that, Step S3 includes the following steps: Step S31: Obtain and construct a multi-dimensional pollution feature database through historical pollution event analysis and river section simulation; Step S32: Calculate the multi-gradient concentration of each type of pollutant using a multi-dimensional pollution feature library to form data on the relationship between concentration and parameter response; Step S33: Based on the concentration and parameter response data, the pollution response process is divided into the initial response period, the rapid change period, and the stable period, and characteristic parameters of each stage are extracted; Step S34: Encode the pollutant features for each pollutant and convert the response features of each pollutant into a 64-bit digital feature code; Step S35: Verify the consistency of digital signature codes under different water temperatures and flow rates, and supplement and improve the multi-dimensional pollution signature database.

8. The comprehensive evaluation method for river ecological health integrating water chemistry and biological indices according to claim 7, characterized in that, Step S33 includes: The point at which the marked parameters begin to deviate from the baseline is taken as the starting point of the initial response period, and the lag time of the initial response is recorded. The period during which the rate of change of the marked parameter reaches its maximum value is the period of rapid change. The peak value and duration of the rate of change are extracted. The period during which the marked parameters tend to stabilize is called the stabilization period. The deviation of the stabilization period parameter values ​​from the baseline is calculated. Key characteristic parameters of the three stages—initial response period, rapid change period, and steady period—are extracted, including initial response delay, peak change rate, and steady-state value offset. Analyze the temporal relationships between different stages to establish a temporal response characteristic map of pollutants; Construct a response phase ratio model for each pollutant and record the percentage of the total response time for each of the three phases.

9. The comprehensive evaluation method for river ecological health integrating hydrochemical and biological indices according to claim 1, characterized in that, Step S3 also includes: When a health index in the dynamic evaluation benchmark framework triggers an early warning, it automatically extracts time-series data of abnormal parameters to form an event slice. Analyze the multidimensional variation characteristics of event slices and calculate the similarity between event slices and features in the multidimensional pollution feature library; Similarity is graded into high match, medium match, and low match, with similarity ≥85% being high match, 75-85% being medium match, and <75% being low match. A high match indicates a single contamination type, a medium match indicates multiple contamination types, and a low match indicates an unknown contamination type.

10. The comprehensive evaluation method for river ecological health integrating hydrochemical and biological indices according to claim 9, characterized in that, Step S3 also includes: Record the timestamps of anomalies occurring at upstream and downstream sections in the three-point monitoring network and calculate the time difference; Obtain the distance data between cross sections and calculate the pollutant propagation speed by combining the time difference; Collect real-time flow velocity data and establish a library of propagation correction coefficients based on pollutant characteristics; The location of the pollution source is calculated by tracing back to the source using a propagation correction coefficient library. Match the location of pollution sources with the river estuary information in the geographic information system; Generate spatial coordinates and positioning accuracy estimates of the pollution source location, and output a map of the investigation area.

Images