Water quality on-line monitoring method and system for large-volume water tank

Abstract

The invention provides a water quality on-line monitoring method and system for a large-volume water tank, and the method comprises the steps: obtaining original data of a plurality of water quality sensors, estimating the water quality state of the whole water tank through a weighted interpolation method, and forming a water quality space distribution baseline; for multi-sensor data, the pollution deviation degree of the sensor data and the water quality space distribution base line is calculated, then according to the pollution deviation degree and a dynamic self-adaptive threshold value, an abnormal pollution event is recognized, and abnormal detection data is obtained; according to the abnormal detection data, utilizing a hydrodynamic model to predict the diffusion trend of pollutants, and evaluating future water quality change; determining an optimal water quality management strategy according to the pollution prediction data in combination with water tank physical parameters and water treatment resource information to obtain a water quality optimization decision scheme; and dynamically adjusting the hydrodynamic model according to the execution result of the water quality optimization decision scheme and the latest monitoring data. According to the method, the defects of an existing method in the aspects of monitoring comprehensiveness, real-time performance and accuracy are overcome.

Description

Technical Field

[0001] The present invention belongs to the technical field of on-line water quality monitoring, and particularly relates to an on-line water quality monitoring method and system for large-volume water tanks. Background Art

[0002] In the water quality management of large-volume water tanks, real-time monitoring is the key to ensuring water quality safety and stability. Traditional water quality monitoring methods mainly rely on fixed sensors and regular manual sampling and detection. Fixed sensors are generally used to measure parameters such as pH value, dissolved oxygen, conductivity, turbidity, ammonia nitrogen concentration, etc., and upload the data to the monitoring system. However, due to the uneven distribution of water quality inside the large-volume water tank, the monitoring data at a single point or a small number of measuring points are difficult to comprehensively reflect the water quality status of the entire water tank, and it is impossible to effectively capture the spatial changes of water quality. In addition, the method of manual sampling and detection not only has high costs and low efficiency, but also due to the possible changes during the transportation and storage of water samples, the timeliness of the data is insufficient, making it difficult to meet the requirements of real-time monitoring. In recent years, some intelligent water quality monitoring systems have introduced wireless sensor networks or Internet of Things technologies to improve the automation degree of data collection, but these methods still face the following main problems: First, the data collected by sensors are easily affected by local noise and environmental interference, resulting in misjudgment of abnormal data points; Second, the water quality state of large-volume water tanks is affected by hydrodynamic characteristics, and the slow movement of water flow may cause local pollutant retention or uneven concentration, while most of the existing water quality prediction models are based on simple statistical methods or empirical formulas, lacking accurate modeling of the spatio-temporal change trend of water quality; Finally, although some water quality monitoring algorithms based on machine learning have been tried and applied, these methods often only rely on historical data for training, have limited detection capabilities for sudden pollution events, cannot accurately identify abnormal pollution sources, and are even less able to effectively predict the pollution diffusion process. Therefore, there is still a lack of a solution for the current water quality monitoring of large-volume water tanks that can comprehensively consider the spatial distribution characteristics of water quality, accurately detect water quality anomalies, and at the same time have strong real-time performance and self-adaptability. Summary of the Invention

[0003] The object of the present invention is to design an on-line water quality monitoring method and system for large-volume water tanks, which overcomes the deficiencies of the existing methods in terms of monitoring comprehensiveness, real-time performance, and accuracy.

[0004] To achieve the above object, in the first aspect of the present invention, an on-line water quality monitoring method for large-volume water tanks, the method includes:

[0005] Obtain the original data of multiple water quality sensors, perform multi-sensor data fusion, and construct a water tank water quality spatial distribution baseline based on the fused data; wherein, the water quality spatial distribution baseline is based on the water tank space grid, and uses a weighted interpolation method to estimate the water quality status of the entire water tank to form a water quality spatial distribution baseline;

[0006] For multi-sensor data, calculate the pollution deviation degree between the sensor data and the water quality spatial distribution baseline, and then identify abnormal pollution events based on the pollution deviation degree and the dynamic adaptive threshold to obtain abnormal detection data; wherein, the abnormal detection data includes the pollution location, the pollution deviation degree, and the pollution type.

[0007] According to the abnormal detection data, use the hydrodynamic model to predict the diffusion trend of pollutants, evaluate the future water quality changes, and form pollution trend prediction data; wherein, the pollution trend prediction data includes the pollutant concentration distribution, the pollution diffusion path, the pollution impact assessment, and the updated water quality spatial distribution baseline.

[0008] According to the pollution prediction data, combine the physical parameters of the water tank and the water treatment resource information to determine the optimal water quality management strategy and obtain the water quality optimization decision plan.

[0009] Dynamically adjust the hydrodynamic model according to the execution result of the water quality optimization decision plan and the latest monitoring data. If the water quality change after implementing the optimization plan continuously deviates from the threshold within the preset time window, adjust the updated water quality spatial distribution baseline to improve the long-term monitoring accuracy of the water quality in the water tank.

[0010] Preferably, the original data includes the pH value, the dissolved oxygen concentration, the conductivity, the turbidity, the water temperature, the position information of the sensor, and the boundary information of the water tank; the boundary information of the water tank is the length, width, and height of the water tank.

[0011] Preferably, constructing the water quality spatial distribution baseline of the water tank based on the fused data specifically includes:

[0012] Divide the entire water tank into multiple small spatial grids, where each grid point (x j , y j , z j ) represents a specific position in the water tank.

[0013] At each grid point (x j , y j , z j ), use the weighted interpolation method to calculate the water quality spatial distribution baseline M baseline , expressed as:

[0014]

[0015] where K represents the number of neighboring sensors used for calculating the interpolation; d j represents the Euclidean distance between the i-th sensor and the grid point (x j , y j , z j ); D raw(x j , y j , z j ) is the original data; among them, the nearest sensor contributes the most, while the distant sensor contributes the least; among them, the water quality spatial distribution baseline M baseline also includes the interpolated complete water quality data.

[0016] Preferably, the pollution deviation degree is constrained based on the non-uniform spatial regularization term; the non-uniform spatial regularization term is obtained by performing spatial error analysis on the original data and the interpolated complete water quality data;

[0017] The dynamic adaptive threshold is generated by weighted summation of the deviation mean, standard deviation, and non-uniform spatial regularization term corresponding to the pollution deviation degree.

[0018] Preferably, the identification of abnormal pollution events further includes: abnormal area detection:

[0019] The determination of the abnormal area is based on the deviation degree of K neighboring points. If the deviation degree of the K neighboring points is greater than the set threshold, it is determined that there is a pollution event in this area, rather than a single-point anomaly, that is, not a single-location anomaly.

[0020] Preferably, the use of the hydrodynamic model to predict the diffusion trend of pollutants specifically includes:

[0021] Let the three-dimensional flow velocity field of the water tank be V(x, y, z), and the diffusion of pollutants is affected by the water flow:

[0022]

[0023] Among them, C(x, y, z, ι) is the change of pollutant concentration with time ι; V(x, y, z) is the water flow velocity vector, indicating the influence of the water flow on pollutants; D is the diffusion coefficient, which determines the diffusion speed of pollutants in the water tank; and respectively represent the gradient and Laplace diffusion term of pollutants, controlling the diffusion rate of pollutants;

[0024] For the hydrodynamic model, based on the boundary difference quantification, a non-uniform boundary condition constraint term is determined to ensure that in the closed environment of the water tank, the diffusion of pollutants will not exceed the physical boundary.

[0025] Among them, according to the diffusion rate D of pollutants and the water flow velocity V(x, y, z), the future distribution of pollutants in the water tank is deduced to form pollution trend prediction data, including:

[0026] According to the pollution source location provided by the anomaly detection data, the initial pollutant concentration distribution is set;

[0027] Using the finite difference method or the lattice Boltzmann method, calculate the diffusion trajectory of pollutants within time T;

[0028] During the diffusion simulation, introduce a non-uniform boundary condition constraint term to ensure that pollutants do not exceed the boundary range of the water tank;

[0029] Output the predicted concentration distribution of pollutants within the next T time.

[0030] Preferably, according to the pollution prediction data, combined with the physical parameters of the water tank and the water treatment resource information, determine the optimal water quality management strategy to obtain a water quality optimization decision-making scheme, including:

[0031] Define a pollution severity index and prioritize the treatment of high-risk areas;

[0032] Taking the minimization of pollution impact and the minimization of resource consumption as the optimization objectives and the upper limit of available resources as the constraint, construct a water quality optimization objective function;

[0033] Among them, the water quality management strategy includes an optimal water change strategy, an optimal chemical dosing plan, and an optimal aeration and oxygenation plan.

[0034] Preferably, the pollution severity index R polluntion is calculated as:

[0035]

[0036] where C(x, y, z, T) is the predicted concentration of pollutants within the next T time; C safe is the safety concentration threshold, and exceeding this value indicates that there is a pollution risk in the water quality; is an indicator function, taking 1 when the pollutant concentration exceeds the standard and 0 otherwise; W(x, y, z) is the regional weight, considering the importance of the polluted area.

[0037] Preferably, according to the execution result of the water quality optimization decision-making scheme and the latest monitoring data, if the prediction error between the pollution trend prediction data of the hydrodynamic model and the actual pollution trend data continuously exceeds the set threshold, then adjust the model parameters of the hydrodynamic model to ensure that the pollution prediction model can adapt to the diffusion patterns of different pollutants during long-term operation and improve the prediction accuracy.

[0038] In a second aspect, the present invention also provides a water quality on-line monitoring system for a large-volume water tank, and the system includes:

[0039] A data fusion module: Obtain the original data of multiple water quality sensors, perform multi-sensor data fusion, and construct a baseline of the water quality spatial distribution of the water tank based on the fused data; wherein, the water quality spatial distribution baseline is based on the spatial grid of the water tank, and uses a weighted interpolation method to estimate the water quality state of the entire water tank to form a water quality spatial distribution baseline;

[0040] Anomaly detection module: For multi-sensor data, calculate the degree of pollution deviation between the sensor data and the water quality spatial distribution baseline, and then identify abnormal pollution events based on the degree of pollution deviation and the dynamic adaptive threshold to obtain anomaly detection data; wherein, the anomaly detection data includes the pollution location, the degree of pollution deviation, and the pollution type;

[0041] Diffusion prediction module: According to the anomaly detection data, use the hydrodynamic model to predict the diffusion trend of pollutants, evaluate the future water quality changes, and form pollution trend prediction data; wherein, the pollution trend prediction data includes the pollutant concentration distribution, the pollution diffusion path, the pollution impact assessment, and the updated water quality spatial distribution baseline;

[0042] Optimization decision-making module: According to the pollution prediction data, combined with the physical parameters of the water tank and the water treatment resource information, determine the optimal water quality management strategy to obtain the water quality optimization decision-making scheme;

[0043] Model update module: Dynamically adjust the hydrodynamic model according to the execution result of the water quality optimization decision-making scheme and the latest monitoring data. If the water quality change after implementing the optimization scheme continuously deviates from the threshold within the preset time window, adjust the updated water quality spatial distribution baseline to improve the long-term monitoring accuracy of the water tank water quality

[0044] The beneficial effects of the present invention are as follows:

[0045] First of all, the present invention constructs a model that can describe the water quality spatial distribution characteristics inside the water tank. Through the dynamic data fusion technology, the measurement data of multiple sensors are optimized to form a global understanding of the water quality status inside the water tank, thus effectively solving the problem that the existing methods cannot reflect the water quality spatial non-uniformity.

[0046] Secondly, the present invention proposes an adaptive anomaly detection mechanism that can combine the internal laws of water quality monitoring data to accurately identify sudden pollution events, reduce misjudgments caused by sensor noise, and improve the sensitivity of anomaly detection, overcoming the defect of insufficient abnormal pollution source identification ability of the existing methods.

[0047] In addition, in order to better predict the future change trend of the water tank water quality, the present invention also introduces a prediction method combined with hydrodynamic characteristics. By analyzing the influence of the water flow inside the water tank on the pollutant diffusion, more accurate water quality prediction results are provided, making up for the deficiency of the existing methods in time series modeling.

[0048] Finally, the present invention optimizes the data processing method in the system architecture, enabling the water quality monitoring system to adaptively adjust the monitoring strategy in real time, improving the intelligent level of monitoring, ensuring the accuracy and timeliness of the water quality monitoring results, and thus providing more reliable technical support for the water quality management of large-volume water tanks. Description of the Drawings

[0049] The present invention will be further described with reference to the accompanying drawings. However, the embodiments in the drawings do not constitute any limitation to the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained according to the following drawings.

[0050] Figure 1 This is a flowchart of the online water quality monitoring method for large-volume water tanks according to the present invention.

[0051] Figure 2 This is a framework diagram of the online water quality monitoring system for large-volume water tanks according to the present invention. Detailed Embodiments

[0052] The embodiments of the present invention will be described in detail below. The examples of the embodiments are shown in the drawings, where the same or similar reference numerals represent the same or similar elements or elements with the same or similar functions throughout. The embodiments described below by referring to the drawings are exemplary and are only used to explain the present invention and should not be construed as a limitation to the present invention.

[0053] In one or more embodiments, as Figure 1 shown, an online water quality monitoring method for large-volume water tanks is disclosed. The method includes the following steps S1 - S5:

[0054] S1. Obtain the original data of multiple water quality sensors, perform multi-sensor data fusion, and construct a baseline of the water quality spatial distribution in the water tank based on the fused data; wherein, the water quality spatial distribution baseline is based on the water tank spatial grid, and the weighted interpolation method is used to estimate the water quality state of the entire water tank to form the water quality spatial distribution baseline.

[0055] Preferably, the original data includes pH value, dissolved oxygen concentration, conductivity, turbidity, water temperature, the position information of the sensors, and the boundary information of the water tank; the boundary information of the water tank is the length, width, and height of the water tank.

[0056] Specifically, the objective of this step is to construct a water quality spatial distribution baseline M inside the entire water tank based on the original data of multiple water quality sensors. baselineThis baseline represents the water quality state of the water tank under normal conditions and provides a benchmark for subsequent anomaly detection (Step 2). If a significant deviation occurs in a certain area of the water tank during subsequent monitoring, the system can determine it as a possible pollution event. Traditional water quality monitoring systems often only focus on the readings of fixed sensors. However, the water quality inside the water tank usually changes continuously and is affected by multiple factors (such as water flow, temperature gradient). The data of a single sensor cannot directly represent the water quality state of the entire water tank. Therefore, the present invention needs to construct a complete three-dimensional water quality distribution model so that the system can still estimate the water quality of areas without sensor coverage.

[0057] Furthermore, the input of this step is the original measurement data of each water quality sensor in the water tank, mainly including the following:

[0058] D raw : The original data of multiple water quality sensors, including: pH value (acidity and alkalinity), dissolved oxygen concentration, conductivity, turbidity, water temperature, and the position information (x i , y i , z i ) of the sensor: Describing the specific three-dimensional coordinates of each sensor in the water tank for constructing the spatial distribution of water quality in the water tank; The boundary information of the water tank: including the length, width, and height of the water tank, for constructing a reasonable water quality distribution model to avoid calculations beyond the physical range.

[0059] Description of data source:

[0060] These data come from fixed water quality sensors distributed at different positions in the water tank, and each sensor reports the measured value regularly (for example, every 5 minutes).

[0061] Since the sensor measurements may be affected by noise interference, such as water flow disturbance, electromagnetic interference, etc., these data may need to be further processed in subsequent steps.

[0062] Preferably, constructing the baseline of the spatial distribution of water quality in the water tank based on the fused data specifically includes:

[0063] Dividing the entire water tank into multiple small spatial grids, where each grid point (X j , y j , z j ) represents a specific position in the water tank;

[0064] At each grid point (x j , y j , z j ), the weighted interpolation method is used to calculate the baseline M of the water quality spatial distribution baseline , expressed as:

[0065]

[0066] Among them, K represents the number of neighboring sensors used for calculating interpolation; d i represents the Euclidean distance between the i-th sensor and the grid point (x j , y j , z j ); D raw (x i , y i , z i ) is the original data; among them, the closest sensor contributes the most, while the distant sensors contribute the least; among them, the water quality spatial distribution baseline M baseline also includes the interpolated complete water quality data.

[0067] Specifically, in order to model the water quality inside the water tank, the present invention first divides the entire water tank into multiple small spatial units (i.e., grids). Each grid point (x j , y j , z j ) represents a specific position in the water tank, and the system needs to estimate the water quality state at this position.

[0068] Grid division method: Regular three-dimensional cubic grids are adopted. For example, a water tank with dimensions of 10m×10m×5m is divided into small cubic units of 1m×1m×1m, and a total of 500 grid points are obtained.

[0069] Function of grid points: These grid points are not sensors, but virtual water quality monitoring points calculated by the system based on sensor data.

[0070] Illustrative example:

[0071] Suppose the sensors in the water tank are unevenly distributed, with more sensors in one area and fewer sensors in another area. Then, when the system divides the grid, it needs to ensure that:

[0072] In the area with dense sensors, the system can directly use the measured data.

[0073] In the area with sparse sensors, the system needs to rely on spatial interpolation technology to supplement the data so that reasonable water quality estimates can be obtained in all areas.

[0074] Due to the limited number of sensors, there are no direct measurement values at some grid points. Therefore, an interpolation method needs to be used to estimate the water quality state at this point based on the data of surrounding sensors. Core idea:

[0075] At each grid point (x j , y j , z j) At this point, the present invention hopes to estimate its water quality status based on the data D of surrounding sensors raw (x i , y i , z i ).

[0076] However, the influence degrees of different sensors on this point are different: the closer the sensor is, the greater the influence; the farther the sensor is, the smaller the influence. Calculation method:

[0077] Adopt the weighted interpolation method, and the weight w i is determined by the distance d between the sensor and the grid point i so that the closer sensors contribute more, while the farther sensors contribute less.

[0078] Among them, w i is calculated by the following formula:

[0079]

[0080] Among them, K: the number of neighboring sensors used for interpolation calculation. d i : the Euclidean distance between the i-th sensor and the grid point (x j , y j , z j ). α: a parameter controlling the attenuation degree, so that the influence of the farther sensors on this point is reduced.

[0081] For example:

[0082] Suppose only one sensor in a certain area A of the water tank measures the pH value to be 7.0, while in another area B, the pH values measured by 3 sensors are 6.8, 7.2, and 7.1 respectively. When the system performs interpolation calculation, it will give higher credibility to the results of area B, rather than simply using the single-point data of A to fill all unknown areas, so as to ensure the rationality of the water quality distribution.

[0083] The final output of this step is: the water quality spatial distribution baseline M baseline ; this baseline describes the stable water quality status of the entire water tank, that is, in the case of no pollution, what the water quality should be like at different positions.

[0084] This baseline will be used in step 2 (anomaly detection) to compare with real-time monitoring data to determine whether there is an anomaly.

[0085] The interpolated complete water quality data D spatial ;

[0086] These data are used for subsequent pollution trend prediction (step 3) and intelligent water quality management (step 4).

[0087] S2. For multi-sensor data, calculate the pollution deviation degree between the sensor data and the water quality spatial distribution baseline, and then identify abnormal pollution events based on the pollution deviation degree and the dynamic adaptive threshold to obtain abnormal detection data; wherein, the abnormal detection data includes the pollution location, the pollution deviation degree, and the pollution type.

[0088] Among them, the goal of this step is based on the water quality spatial distribution baseline M baseline , to detect abnormal pollution events in the water tank in real time and ensure that the detected abnormalities are caused by actual pollution, rather than normal changes in sensor errors or environmental fluctuations. By comparing the current measurement data with M baseline , the system can accurately identify pollution abnormalities, calculate the spatial range of the abnormal area, and provide key data for the follow-up.

[0089] The input data of this step comes from the output of step 1 and includes:

[0090] M baseline : The water quality spatial distribution baseline, representing the water quality state of the water tank under normal conditions.

[0091] D spatial : The complete water quality data after spatial interpolation, enabling the detection to cover not only the locations of the sensors but also the entire water tank area.

[0092] D raw : The real-time measurement data of the water tank sensors, used to compare with M baseline to judge the abnormal state.

[0093] Sensor position information (x i , y i , z i ): Used to locate the specific position of the abnormal point in three-dimensional space.

[0094] Preferably, the pollution deviation degree is constrained based on the inhomogeneous spatial regularization term; the inhomogeneous spatial regularization term is obtained by performing spatial error analysis on the original data and the interpolated complete water quality data;

[0095] The dynamic adaptive threshold is generated by performing weighted summation on the deviation mean, standard deviation, and inhomogeneous spatial regularization term corresponding to the pollution deviation degree.

[0096] Preferably, the identification of abnormal pollution events further includes: Abnormal area detection:

[0097] The determination of the abnormal area is based on the deviation degrees of K neighboring points. If the deviation degrees of the K neighboring points are greater than the set threshold, it is determined that there is a pollution event in this area, rather than a single-point abnormality, that is, not a single-position abnormality.

[0098] Specifically, (1) Spatial deviation calculation:

[0099] Traditional anomaly detection methods usually rely on fixed thresholds (such as θ). However, the water quality inside the water tank is dynamically changing, and fixed thresholds are prone to false alarms or missed alarms. This step adopts an improved spatial deviation calculation method to ensure detection accuracy.

[0100] Calculate the sensor measurement value D raw (x i ,y i ,z i ) and the deviation degree from the baseline M baseline :

[0101] ΔW(x i ,y i ,z i ) = |D rew (x i ,y i ,z i | - M bseline (x i ,y i ,z i )|

[0102] If ΔW exceeds a certain threshold, there may be abnormal pollution.

[0103] Introduce an inhomogeneous spatial regularization term through spatial error analysis

[0104]

[0105] Among them, γ i is the regularization weight, representing the reliability of the sensor signal (obtained from historical error statistics). This term ensures that spatial inhomogeneity is considered during anomaly detection and reduces misjudgments caused by local sensor failures.

[0106] (2) Adaptive anomaly threshold calculation:

[0107] Due to the dynamically changing water tank environment, fixed thresholds are not applicable to all regions. Therefore, this step adopts dynamic adaptive threshold calculation:

[0108]

[0109] Among them, μ SW : The mean deviation in the last T rounds of detections. o ΔW : The standard deviation of the deviation, reflecting the water quality fluctuation range. η: The spatial regularization adjustment parameter to prevent false alarms of local anomalies.

[0110] The present invention combines spatial regularization to ensure that the threshold is higher in areas with large water quality fluctuations and lower in stable areas, improving the detection accuracy.

[0111] (3) Abnormal area determination:

[0112] Pollution events usually do not affect a single sensor only, but form a certain spatial diffusivity. Therefore, it is necessary to detect whether the abnormality has spatial continuity.

[0113] Calculate the deviation degree of K neighboring points around the abnormal point:

[0114]

[0115] If S anomaly is greater than the set threshold, it is determined that there is a pollution event in this area, rather than a sensor error.

[0116] Illustrate with an example:

[0117] If 8 out of 10 sensors in a certain area exceed the threshold, it can be determined that there is pollution in this area, rather than a single-point abnormality.

[0118] Finally, output data D ano maly : including information on abnormal pollution points, including: pollution location (x a , y a , z a ), pollution deviation degree ΔW(x a , y a , z a ), pollution type estimation, updated Adjust the baseline appropriately to update it with the long-term change of water quality.

[0119] S3. According to the abnormal detection data, use the hydrodynamic model to predict the diffusion trend of pollutants, evaluate the future water quality change, and form pollution trend prediction data; wherein, the pollution trend prediction data includes pollutant concentration distribution, pollution diffusion path, pollution impact assessment, and updated water quality spatial distribution baseline.

[0120] Preferably, the use of the hydrodynamic model to predict the diffusion trend of pollutants specifically includes:

[0121] Set the three-dimensional flow velocity field of the water tank as V(x, y, z), and the diffusion of pollutants is affected by the water flow:

[0122]

[0123] Among them, C(x, y, z, ι) is the change of pollutant concentration with time ι; V(x, y, z) is the water flow velocity vector, indicating the influence of water flow on pollutants; D is the diffusion coefficient, which determines the diffusion speed of pollutants in the water tank; and respectively represent the gradient and Laplace diffusion term of pollutants, controlling the diffusion rate of pollutants;

[0124] For the hydrodynamic model, a non-uniform boundary condition constraint term is determined based on boundary difference quantification to ensure that the diffusion of pollutants does not exceed the physical boundary in the closed environment of the water tank.

[0125] Among them, according to the diffusion rate D of pollutants and the water flow velocity V(x, y, z), the future distribution of pollutants in the water tank is estimated to form pollution trend prediction data, including:

[0126] Set the initial pollutant concentration distribution according to the pollution source location provided by the anomaly detection data;

[0127] Use the finite difference method or the lattice Boltzmann method to calculate the diffusion trajectory of pollutants within time T;

[0128] During the diffusion simulation process, introduce a non-uniform boundary condition constraint term to ensure that pollutants do not exceed the boundary range of the water tank;

[0129] Output the predicted concentration distribution of pollutants within the future time T.

[0130] Specifically, the goal of this step is based on the anomaly detection data D anomaly , combined with the hydrodynamic characteristics inside the water tank, to predict the diffusion trend of pollutants, and then evaluate the future changes in water quality. Since the water quality state in the water tank is affected not only by the pollution source location but also by factors such as water flow, temperature gradient, and diffusion effect, it is impossible to accurately predict the pollution range relying solely on the anomaly point information. This step forms a spatio-temporal prediction model for pollutant diffusion through hydrodynamic modeling + diffusion law calculation, and provides a basis for subsequent water quality optimization decisions.

[0131] Among them, the input data for this step comes from the output of step 2, including:

[0132] D anomaly : Anomaly detection data, including the pollution location (x a , y a , z a ), the pollution deviation degree ΔW(x a , y a , z a ), pollution type estimation, etc.

[0133] M baselne : The baseline of water quality spatial distribution, used to refer to the water quality distribution under normal conditions.

[0134] Hydrodynamic parameters: including the flow velocity field V(x, y, z), water flow direction, temperature gradient, etc. inside the water tank, which are obtained by sensor measurement or experimental calibration.

[0135] The physical boundaries of the water tank: used to calculate the hydrodynamic properties and ensure that the simulation process conforms to the actual conditions of the water tank.

[0136] Specifically, (1) Hydrodynamic modeling:

[0137] The diffusion of pollutants in the water tank is not only controlled by random diffusion, but also affected by water flow and convection. Therefore, the present invention adopts an improved hydrodynamic prediction model to calculate the diffusion path of pollutants.

[0138] Among them, the traditional hydrodynamic model is usually used in open water. The innovation of this patent design is to propose non-uniform boundary condition constraints for closed water tank environments, and to quantify the non-uniform boundary condition constraints based on boundary differences:

[0139]

[0140] Where, B: the set of boundary points in the water tank. C b : Pollutant concentration at the border of the water tank. C adj : Pollutant concentration in the adjacent area. κ b : Boundary conditions adjust parameters to ensure that pollutants do not diffuse infinitely but are constrained by the closed boundaries of the water tank.

[0141] This constraint ensures that the diffusion of pollutants does not exceed the physical boundary in the closed environment of the water tank, while avoiding errors caused by boundary effects.

[0142] (2) Spatiotemporal pollution prediction:

[0143] According to the diffusion rate D of pollutants and the water flow velocity V (x, y, z), the distribution of pollutants in the water tank in the future is estimated to form the pollution trend prediction data D forecast .

[0144] Calculation steps:

[0145] Initialize the pollutant diffusion state: According to D anomaly The pollution source location (x a ,y a , z a ), set the initial pollutant concentration distribution.

[0146] Based on the solution of hydrodynamic equations: using the finite difference method or the lattice Boltzmann method (existing counting), the diffusion trajectory of pollutants within time T is calculated.

[0147] Combined boundary constraints: In the diffusion simulation process, introduce Make sure contaminants do not extend beyond the boundaries of the tank.

[0148] Generate pollution prediction data: Output the predicted concentration distribution of pollutants in the future T time.

[0149] For example, if an abnormal pollution source is detected in a certain area, with its pH value dropping sharply by 0.8 units and the water flow direction shifting 10° to the southeast, the system will calculate the diffusion path of the pollutant within the next 30 minutes and predict the changing trend of the pollutant concentration at different time points.

[0150] Final output: D forecast : Pollution trend prediction data, including: the predicted values of the pollutant concentration distribution C(x, y, z, ι) within the next T time. Pollution diffusion path: the diffusion range of the pollutant over time. Pollution impact assessment: whether the pollutant concentration will exceed the safety threshold and whether it will affect a specific area. Updated Adjust the baseline model appropriately to better adapt to future water quality changes.

[0151] S4. Based on the pollution prediction data, combined with the physical parameters of the water tank and the water treatment resource information, determine the optimal water quality management strategy to obtain a water quality optimization decision plan.

[0152] Preferably, the step of determining the optimal water quality management strategy based on the pollution prediction data, combined with the physical parameters of the water tank and the water treatment resource information, to obtain a water quality optimization decision plan includes:

[0153] Define a pollution severity index and prioritize the treatment of high-risk areas;

[0154] Taking minimizing pollution impact and minimizing resource consumption as the optimization objectives and the upper limit of available resources as the constraint, construct a water quality optimization objective function;

[0155] Among them, the water quality management strategy includes the optimal water change strategy, the optimal chemical dosing plan, and the optimal aeration and oxygenation plan.

[0156] Specifically, the objective of this step is based on the pollution trend data D predicted in step 3 forecast , formulate the optimal water quality management strategy to reduce the pollution impact and maintain the long-term stability of the water quality in the water tank. This strategy should not only deal with the currently detected pollution but also predict future pollution trends and take measures in advance to achieve the optimal water quality management plan with the lowest cost and minimum resource consumption.

[0157] Traditional water quality optimization methods usually adopt fixed rules, such as regular water changes or fixed-dose chemical dosing. These methods have the following problems:

[0158] Unable to adapt to the dynamic changes of pollution, which may lead to excessive or insufficient water treatment solutions.

[0159] Do not consider the pollution diffusion trend and can only respond passively rather than intervene in advance.

[0160] Ignoring the hydrodynamic characteristics in the water tank may lead to uneven distribution of the agent or insufficient dilution of the pollution.

[0161] This step proposes an intelligent water quality optimization strategy, based on pollution impact assessment + resource optimization scheduling + adaptive water quality management, to ensure the scientificity and feasibility of the water tank water quality optimization plan.

[0162] Specifically, the input data for this step comes from the output of step 3, including:

[0163] D forecast : Pollution trend prediction data, describing the change of pollutant concentration C(x, y, z, ι) in the next T time.

[0164] Physical parameters of the water tank: including the water tank volume v tank 、Inlet and outlet water flow rates Q in ,Q out 、Hydrodynamic model V(x, y, z), etc.

[0165] Water treatment resource information: including available water treatment methods (water replacement, chemical agent dosing, aeration and oxygenation, etc.), resource consumption constraints (maximum water replacement volume, chemical agent inventory, etc.).

[0166] Furthermore, different pollutants have different impacts on water quality. It is necessary to calculate the impact degree of pollutants on the overall water tank water quality to ensure that the optimization plan gives priority to treating high-risk areas. Define the pollution severity index:

[0167]

[0168] Among them, C(x, y, z, T): Predicted concentration of pollutants in the next T time. C safe : Safety concentration threshold. Exceeding this value indicates that there is a pollution risk in the water quality. Indicator function, taking 1 when the pollutant concentration exceeds the standard, otherwise taking 0. W(x, y, z): Regional weight, considering the importance of the polluted area (such as the area near the water tank outlet has a greater weight).

[0169] It can be understood that traditional methods usually only focus on the current concentration of pollutants, while this solution calculates R through future pollution trends pollwtion to ensure that the optimization plan has forward-looking. The weighted factor W(x, y, z) is adopted to give priority to optimizing key areas and reduce the impact of pollution on important parts, such as the drinking water outlet or the key area of industrial water.

[0170] Furthermore, the water quality optimization objective function: while optimizing the water quality, ensure the minimization of resource use and avoid unnecessary water treatment operations.

[0171] Objective 1: Minimize the pollution impact

[0172]

[0173] Among them, C target : the target water quality concentration, usually set as the safety threshold C safe .

[0174] Goal 2: Minimize resource consumption

[0175] ∑R pesource ≤R max

[0176] Among them: R resouce : the resource consumption of different water treatment methods (such as the cost of water replacement, the dosage of chemicals). R max : the upper limit of available resources.

[0177] Optimization strategy:

[0178] Water replacement optimization: If the concentration in the polluted area is high and the pollutants can be diluted by water replacement, then calculate the optimal water replacement volume v flush , ensuring that the pollutants are diluted with the least water loss.

[0179] Chemical dosage optimization: If the pollutants are chemically degradable pollutants, then calculate the best chemical dosage to ensure that the pollutant concentration is lower than the safety threshold after dosing.

[0180] Aeration optimization: If the pollutants are dissolved pollutants (such as low dissolved oxygen areas), then calculate the best aeration intensity to enhance the self-purification ability of the water body.

[0181] S5. Dynamically adjust the hydrodynamic model according to the execution result of the water quality optimization decision plan and the latest monitoring data. If the water quality change after implementing the optimization plan continuously deviates from the threshold within the preset time window, then adjust the updated water quality spatial distribution baseline to improve the long-term monitoring accuracy of the water quality in the water tank.

[0182] Preferably, according to the execution result of the water quality optimization decision plan and the latest monitoring data, if the prediction error between the pollution trend prediction data and the actual pollution trend data of the hydrodynamic model continuously exceeds the set threshold, then adjust the model parameters of the hydrodynamic model to ensure that the pollution prediction model can adapt to the diffusion patterns of different pollutants during long-term operation and improve the prediction accuracy.

[0183] Specifically, the goal of this step is based on the water quality optimization execution result D deoision and the latest water quality monitoring data to dynamically update the water quality monitoring model, enabling the system to adapt to environmental changes in the long term and improving the monitoring accuracy and water quality prediction ability. Since the water quality in the water tank is affected by external environment, influent water quality changes, seasonal factors, etc., after long-term operation, the original water quality spatial distribution baseline Mbaseline and the pollution prediction model M prediot may deviate from the actual water quality status. Therefore, it is necessary to design an adaptive update mechanism to ensure that the model can be continuously optimized.

[0184] Among them, the input data for this step comes from the output of the previous step, including:

[0185] D decision : The water quality optimization plan generated in step 4, including the implementation of measures such as water replacement and chemical dosing.

[0186] The latest water quality monitoring data, that is, the real-time water quality data measured by the sensor after the system executes the optimization strategy.

[0187] M baseline : The current baseline of the water quality spatial distribution, which needs to be adjusted according to the new data.

[0188] M predict : The pollution diffusion prediction model, which needs to correct the parameters according to the optimization execution situation.

[0189] Furthermore, the update of the baseline model M baseline :

[0190] Core idea: The baseline of the water quality spatial distribution M baseline needs to be updated over time to adapt to long-term water quality changes and ensure that the reference values for anomaly detection and pollution prediction are always accurate.

[0191] Update strategy: Calculate the water quality change ΔW after executing the optimization plan opt :

[0192]

[0193] If ΔW opt continually deviates from the threshold θ within a certain time window update , then adjust the baseline:

[0194]

[0195] where β is the weight factor of historical data, which controls the influence of new data on the baseline. This update strategy ensures that the baseline model can be adjusted with the slow changes in water quality without being disturbed by short-term noise.

[0196] It can be understood that traditional water quality baselines are usually based on long-term historical means, while this solution can be dynamically adjusted according to the optimization execution results to always be close to the latest water quality status.

[0197] By setting the adaptive threshold θ updare, avoid the instability caused by frequent model updates and improve the long-term monitoring accuracy.

[0198] Furthermore, for the update of the pollution prediction model M predlat :

[0199] Core idea: The pollution diffusion prediction model needs to be corrected according to the actual pollutant diffusion situation to ensure that the prediction results are consistent with the actual observation values.

[0200] Update strategy: Calculate the prediction error ε predlat :

[0201]

[0202] If ε predlat continuously exceeds the set threshold, then adjust the model parameter θ model :

[0203]

[0204] where η is the learning rate, which controls the update speed of the model parameters. This update method ensures that the pollution prediction model can adapt to the diffusion patterns of different pollutants during long-term operation and improves the prediction accuracy.

[0205] It can be understood that traditional methods usually adopt static parameters, while this solution adjusts θ model through error feedback, continuously optimizing the prediction model to adapt to environmental changes. Combining with the reinforcement learning framework, the model can be continuously adjusted during long-term operation, enabling the system to always maintain high-precision prediction capabilities.

[0206] Finally, the output data:

[0207] The updated new water quality spatial distribution baseline makes anomaly detection and water quality prediction more accurate.

[0208] The updated new pollution diffusion prediction model improves the reliability of future water quality prediction.

[0209] The optimized monitoring system S adaptive : The entire water quality monitoring system has self-adaptive capabilities during long-term operation, improving the stability of long-term monitoring.

[0210] In one or more embodiments, as Figure 2 shown, an online water quality monitoring system for large-volume water tanks is disclosed, and the system includes:

[0211] Data fusion module 100: Obtain the raw data of multiple water quality sensors, perform multi-sensor data fusion, and construct a baseline for the spatial distribution of the water quality in the water tank based on the fused data; wherein, the baseline for the spatial distribution of the water quality is based on the spatial grid of the water tank, and the weighted interpolation method is used to estimate the water quality state of the entire water tank to form the baseline for the spatial distribution of the water quality.

[0212] Abnormal detection module 200: For multi-sensor data, calculate the degree of pollution deviation between the sensor data and the baseline for the spatial distribution of the water quality, and then identify abnormal pollution events based on the degree of pollution deviation and the dynamic adaptive threshold to obtain abnormal detection data; wherein, the abnormal detection data includes the pollution location, the degree of pollution deviation, and the pollution type.

[0213] Diffusion prediction module 300: According to the abnormal detection data, use the hydrodynamic model to predict the diffusion trend of pollutants, evaluate the future water quality changes, and form pollution trend prediction data; wherein, the pollution trend prediction data includes the pollutant concentration distribution, the pollution diffusion path, the pollution impact assessment, and the updated baseline for the spatial distribution of the water quality.

[0214] Optimization decision-making module 400: According to the pollution prediction data, combined with the physical parameters of the water tank and the water treatment resource information, determine the optimal water quality management strategy to obtain a water quality optimization decision-making plan.

[0215] Model update module 500: Dynamically adjust the hydrodynamic model according to the execution result of the water quality optimization decision-making plan and the latest monitoring data. If the water quality change after the execution of the optimization plan continuously deviates from the threshold within the preset time window, then adjust the updated baseline for the spatial distribution of the water quality to improve the long-term monitoring accuracy of the water quality in the water tank.

[0216] It should be noted that the specific working process of the water quality online monitoring system for large-volume water tanks provided in the embodiments of the present invention is the same as the process of the water quality online monitoring method for large-volume water tanks described in the above embodiments, and will not be elaborated here.

[0217] Compared with the prior art, the water quality online monitoring system for large-volume water tanks provided by the embodiments of the present invention acquires the original data of multiple water quality sensors, performs multi-sensor data fusion, and constructs a baseline for the spatial distribution of the water quality in the water tank based on the fused data; wherein, the baseline for the spatial distribution of the water quality is based on the spatial grid of the water tank, and the weighted interpolation method is used to estimate the water quality state of the entire water tank to form the baseline for the spatial distribution of the water quality; for the multi-sensor data, calculate the pollution deviation degree between the sensor data and the baseline for the spatial distribution of the water quality, and then identify abnormal pollution events according to the pollution deviation degree and the dynamic adaptive threshold to obtain abnormal detection data; wherein, the abnormal detection data includes the pollution location, the pollution deviation degree, and the pollution type; according to the abnormal detection data, use the hydrodynamic model to predict the diffusion trend of the pollutant, evaluate the future water quality change, and form pollution trend prediction data; wherein, the pollution trend prediction data includes the pollutant concentration distribution, the pollution diffusion path, the pollution impact assessment, and the updated baseline for the spatial distribution of the water quality; according to the pollution prediction data, combine the physical parameters of the water tank and the water treatment resource information to determine the optimal water quality management strategy to obtain a water quality optimization decision-making plan; dynamically adjust the hydrodynamic model according to the execution result of the water quality optimization decision-making plan and the latest monitoring data. If the water quality change after the implementation of the optimization plan continuously deviates from the threshold within the preset time window, adjust the updated baseline for the spatial distribution of the water quality to improve the long-term monitoring accuracy of the water quality in the water tank.

[0218] The embodiments of the present invention also provide a water quality online monitoring device for a large-volume water tank, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the steps in the embodiments of the water quality online monitoring method for a large-volume water tank as described above are implemented, such as Figure 1 the steps S1 to S5 described therein; or, when the processor executes the computer program, the functions of each module in the above system embodiments are implemented.

[0219] Exemplarily, the computer program can be divided into one or more modules. The one or more modules are stored in the memory and executed by the processor to complete the present invention. The one or more modules can be a series of computer program instruction segments capable of performing specific functions, and these instruction segments are used to describe the execution process of the computer program in the water quality online monitoring device for a large-volume water tank.

[0220] The water quality online monitoring device for a large-volume water tank can be a computing device such as a desktop computer, a notebook, a palm computer, and a cloud server. The water quality online monitoring device for a large-volume water tank may include, but is not limited to, a processor and a memory. Those skilled in the art can understand that the water quality online monitoring device for a large-volume water tank may further include input / output devices, network access devices, a bus, etc.

[0221] The processor may be a central processing unit (CPU), or may also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The processor is the control center of the online water quality monitoring device for large-volume water tanks, and connects all parts of the online water quality monitoring device for large-volume water tanks through various interfaces and lines.

[0222] The memory can be used to store the computer programs and / or modules. By running or executing the computer programs and / or modules stored in the memory, and by calling the data stored in the memory, the processor realizes various functions of the online water quality monitoring device for large-volume water tanks. The memory mainly includes a program storage area and a data storage area. Among them, the program storage area can store an operating system, application programs required for at least one function, etc.; the data storage area can store data created according to the operation of the air-conditioning controller, etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disks, memory, plug-in hard disks, smart media cards (SMCs), secure digital (SD) cards, flash cards, at least one magnetic disk storage device, flash memory devices, or other volatile solid-state storage devices.

[0223] Among them, if the modules integrated in the on-line water quality monitoring device for large-volume water tanks are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, to implement all or part of the processes in the above-described embodiment methods of the present invention, it can also be completed by a computer program instructing relevant hardware. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, the steps of the above-described various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0224] Those of ordinary skill in the art can understand that to implement all or part of the processes in the above-described embodiment methods, it can be completed by a computer program instructing relevant hardware. The program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the above-described method embodiments. Among them, the storage medium can be a magnetic disk, optical disc, read-only memory (ROM, Read-Only Memory), or random access memory (RAM, Random Access Memory), etc.

[0225] The above is the preferred embodiment of the present invention. It should be noted that for those of ordinary skill in the art of this technology, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements are also regarded as the protection scope of the present invention.

Claims

1. An on-line water quality monitoring method for a large-capacity water tank, characterized in that, The method comprises: Obtaining raw data from multiple water quality sensors, performing multi-sensor data fusion, and constructing a water quality spatial distribution baseline for the water tank based on the fused data; wherein the water quality spatial distribution baseline is based on the water tank spatial grid, and a weighted interpolation method is used to estimate the water quality status of the entire water tank to form a water quality spatial distribution baseline; For multi-sensor data, the pollution deviation degree of the sensor data from the water quality spatial distribution baseline is calculated, and then the abnormal pollution events are identified according to the pollution deviation degree and the dynamic adaptive threshold to obtain abnormal detection data; wherein the abnormal detection data includes the pollution location, pollution deviation degree and pollution type; Based on the abnormal detection data, a hydrodynamic model is used to predict the diffusion trend of pollutants, evaluate future water quality changes, and form pollution trend prediction data; wherein the pollution trend prediction data includes pollutant concentration distribution, pollution diffusion path, pollution impact assessment, and updated water quality spatial distribution baseline; According to the pollution prediction data, combined with the physical parameters of the water tank and the water treatment resource information, the optimal water quality management strategy is determined to obtain the water quality optimization decision-making plan; The hydrodynamic model is dynamically adjusted according to the execution results of the water quality optimization decision-making plan and the latest monitoring data. If the water quality changes after the execution of the optimization plan continue to deviate from the threshold within the preset time window, the updated water quality spatial distribution baseline is adjusted to improve the water quality of the water tank and the long-term monitoring accuracy.

2. The on-line water quality monitoring method for a large-volume water tank according to claim 1, characterized in that, The raw data includes pH value, dissolved oxygen concentration, conductivity, turbidity, water temperature, sensor location information and water tank boundary information; the water tank boundary information is the length, width and height of the water tank.

3. The on-line water quality monitoring method for large-capacity water tanks according to claim 1, characterized in that, The method of constructing a water tank water quality spatial distribution baseline based on the fused data specifically includes: Divide the entire water tank into multiple small spatial grids, where each grid point (x j , y j , z j ) represents a specific position in the water tank; At each grid point (x j , y j , z j ), the weighted interpolation method is used to calculate the water quality spatial distribution baseline M baseline , which is expressed as: where K represents the number of neighboring sensors used for calculating the interpolation; d i represents the Euclidean distance between the i-th sensor and the grid point (x j , y j , z j ); D raw (x i , y i , z i ) is the original data; among them, the closest sensor contributes the most, while the distant sensors contribute the least; among them, the water quality spatial distribution baseline M baseline also includes the interpolated complete water quality data.

4. The on-line water quality monitoring method for large-volume water tanks according to claim 1, characterized in that, The pollution deviation degree is constrained based on the non-uniform space regularization term; the non-uniform space regularization term is obtained by performing spatial error analysis based on the original data and the interpolated complete water quality data; The dynamic adaptive threshold is generated by weighted summing of the deviation mean, standard deviation and non-uniform space regularization term corresponding to the pollution deviation degree.

5. The online water quality monitoring method for a large-volume water tank according to claim 4, characterized in that, The identification of abnormal pollution events also includes: abnormal area detection: The determination of the abnormal area is based on the degree of deviation of K neighboring points. If the degree of deviation of K neighboring points is greater than a set threshold, it is determined that there is a pollution event in the area, rather than a single point abnormality, that is, not a single position abnormality.

6. The online water quality monitoring method for a large-volume water tank according to claim 4, characterized in that The method of using a hydrodynamic model to predict the diffusion trend of pollutants specifically includes: Assume that the three-dimensional velocity field of the water tank is V(x, y, z), and the diffusion of pollutants is affected by the water flow: where C(x, y, z, l) is the change in pollutant concentration over time ι; V(x, y, z) is the water flow velocity vector, representing the impact of water flow on pollutants; D is the diffusion coefficient, determining the diffusion rate of pollutants in the water tank; and represent the gradient and Laplacian diffusion terms of pollutants respectively, controlling the diffusion rate of pollutants; For the hydrodynamic model, the non-uniform boundary condition constraint items are quantified based on the boundary difference to ensure that the diffusion of pollutants does not exceed the physical boundary in the closed environment of the water tank; Among them, according to the diffusion rate D of pollutants and the water flow velocity v (x, y, z), the future distribution of pollutants in the water tank is estimated to form pollution trend prediction data, including: According to the pollution source location provided by the anomaly detection data, the initial pollutant concentration distribution is set; Use the finite difference method or the lattice Boltzmann method to calculate the diffusion trajectory of pollutants within time T; During the diffusion simulation process, a non-uniform boundary condition constraint term is introduced to ensure that pollutants do not exceed the boundary range of the water tank; Output the predicted concentration distribution of pollutants within the next T time.

7. The on-line water quality monitoring method for a large-volume water tank according to claim 1, characterized in that According to the pollution prediction data, combined with the physical parameters of the water tank and the water treatment resource information, determine the optimal water quality management strategy to obtain the water quality optimization decision-making scheme, including: Define the pollution severity index and prioritize the treatment of high-risk areas; Taking the minimization of pollution impact and the minimization of resource consumption as the optimization objectives and the upper limit of available resources as the constraint, construct the water quality optimization objective function; Among them, the water quality management strategy includes the optimal water replacement strategy, the optimal chemical dosing plan, and the optimal aeration and oxygenation plan.

8. The on-line water quality monitoring method for a large-capacity water tank according to claim 7, characterized in that, The pollution severity index R pollution is calculated as: where C(x, y, z, T) is the predicted concentration of pollutants within the future time T; C safe is the safety concentration threshold, and exceeding this value indicates a pollution risk in the water quality; is the indicator function, taking 1 when the pollutant concentration exceeds the standard and 0 otherwise; W(x, y, z) is the regional weight, considering the importance of the polluted area.

9. The on-line water quality monitoring method for large-volume water tanks according to claim 1, characterized in that According to the execution result of the water quality optimization decision-making scheme and the latest monitoring data, if the prediction error between the pollution trend prediction data of the hydrodynamic model and the actual pollution trend data continuously exceeds the set threshold, then adjust the model parameters of the hydrodynamic model to ensure that the pollution prediction model can adapt to the diffusion patterns of different pollutants during long-term operation and improve the prediction accuracy.

10. An on-line water quality monitoring system for a large-volume water tank, characterized in that, The system includes: Data fusion module: Obtain the original data of multiple water quality sensors, perform multi-sensor data fusion, and construct a baseline for the spatial distribution of the water tank water quality based on the fused data; among them, the water quality spatial distribution baseline is based on the water tank spatial grid, and the weighted interpolation method is used to estimate the water quality status of the entire water tank to form the water quality spatial distribution baseline; Abnormal detection module: For multi-sensor data, calculate the pollution deviation degree between the sensor data and the water quality spatial distribution baseline, and then identify abnormal pollution events according to the pollution deviation degree and the dynamic adaptive threshold to obtain abnormal detection data; among them, the abnormal detection data includes the pollution location, the pollution deviation degree, and the pollution type; Diffusion prediction module: According to the abnormal detection data, use the hydrodynamic model to predict the diffusion trend of pollutants, evaluate the future water quality changes, and form pollution trend prediction data; among them, the pollution trend prediction data includes the pollutant concentration distribution, the pollution diffusion path, the pollution impact assessment, and the updated water quality spatial distribution baseline; Optimization decision module: According to the pollution prediction data, combined with the physical parameters of the water tank and the water treatment resource information, determine the optimal water quality management strategy to obtain the water quality optimization decision-making scheme; Model update module: Dynamically adjust the hydrodynamic model according to the execution result of the water quality optimization decision-making scheme and the latest monitoring data. If the water quality change after implementing the optimization plan continuously deviates from the threshold within the preset time window, then adjust the updated water quality spatial distribution baseline to improve the water quality of the water tank and the long-term monitoring accuracy.

Images