Water quality monitoring system and method for environmental protection

Abstract

The invention discloses a water quality monitoring system and method for environmental protection, relates to the technical field of water quality monitoring, and combines water quality data and environmental data to form multi-dimensional data fusion analysis and avoid the limitation of single parameter analysis, so that the system has global water quality change judgment capability. The obtained water quality data and environment data are combined and fitted into the original data set W, and optimization is carried out on the data level, so that the data have higher consistency and integrity, the influence of missing data on subsequent analysis is reduced, and the data stability is improved. By monitoring the complexity of the algae community structure, the eutrophication degree of the water body can be accurately evaluated, and an accurate regulation and control basis is provided for water quality treatment; a high-precision sensor and a remote sensing technology are adopted, automatic, remote and continuous collection of water quality data is achieved, manual intervention is reduced, the monitoring cost is reduced, the efficiency and the real-time performance of data acquisition are improved, and a foundation is laid for intelligent water affair and automatic environment management.

Description

Technical Field

[0001] This invention relates to the field of water quality monitoring technology, specifically to a water quality monitoring system and method for environmental protection. Background Technology

[0002] Water environment monitoring falls under the fields of environmental science and ecological protection, encompassing multiple aspects such as water resource management, water pollution control, and aquatic ecosystem protection. With accelerated industrialization and urbanization, water quality safety issues are becoming increasingly serious, making the monitoring of water pollution sources, pollutant concentrations, and their changing trends particularly important. Against this backdrop, water quality monitoring technologies have gradually developed, evolving from traditional manual sampling and analysis to automated and intelligent water quality monitoring systems.

[0003] Existing water quality monitoring systems still primarily rely on alarm mechanisms based on fixed thresholds, making them ill-suited to complex and ever-changing aquatic environments and unable to provide accurate predictions and early warnings. While current water quality monitoring systems can collect water quality parameters in real time, they still have many limitations, mainly in insufficient data analysis capabilities and a lack of predictive ability. Existing systems typically employ traditional threshold-based methods, triggering an alarm only when water quality parameters exceed preset safety thresholds.

[0004] Traditional water quality monitoring systems rely excessively on static rules and lack intelligent data analysis and predictive capabilities. The core of existing water quality monitoring systems depends on fixed threshold settings, which ignores the dynamic changes in water bodies and fails to provide in-depth analysis of historical data and environmental factors. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a water quality monitoring system and method for environmental protection, solving the problems mentioned in the background section.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solution: a water quality monitoring system for environmental protection, comprising a data acquisition module, a data preprocessing module, a feature engineering module, a model training and dynamic optimization module, a water quality prediction module, and a decision support module;

[0007] The data acquisition module collects water quality and environmental data through sensors and fits them into a raw data set W;

[0008] The data preprocessing module performs anomaly detection, data interpolation, and standardization on the collected raw data set W to obtain the water quality dataset SZW;

[0009] The feature engineering module extracts feature variables from the acquired water quality dataset SZW and uses principal component analysis to reduce the dimensionality of the features to obtain the water quality feature set FZ.

[0010] The model training and dynamic optimization module uses the acquired water quality feature set FZ to predict water quality and obtain the water quality monitoring result WQS by using a long short time memory network model.

[0011] The water quality prediction module uses a self-attention mechanism model to perform time series prediction on the water quality monitoring results WQS, predicts the future trend of water quality, and obtains the water quality score WQS(t+p) at time t+p.

[0012] The decision support module calculates the water quality deterioration risk index WQI based on the water quality score WQS(t+p) at time t+p and generates decision recommendations.

[0013] Preferably, the data acquisition module includes a water quality data acquisition unit and an environmental data acquisition unit;

[0014] The water quality data acquisition unit collects water quality data, including microturbulence intensity Sm, sediment suspension rate Sr, and organic matter degradation rate So, through a microturbulence intensity sensor, a suspended particle optical sensor, and a dissolved oxygen and organic matter analyzer.

[0015] The microturbulence intensity Sm was collected by a microturbulence intensity sensor, the sediment suspension rate Sr was obtained by a suspended particle optical sensor, and the organic matter degradation rate So was collected by a dissolved oxygen and organic matter analyzer.

[0016] The microturbulence intensity Sm is obtained by the following formula:

[0017] ;

[0018] In the formula, xus, yus and zus represent the instantaneous flow velocities of the water in the x, y and z directions, respectively, and xup, yup and zup represent the average flow velocities of the water in the x, y and z directions, respectively.

[0019] Sediment suspension rate Sr is obtained using the following formula:

[0020] ;

[0021] In the formula, Ct represents the concentration of suspended particles in the collected water sample, and Co represents the concentration of suspended particles in the background water body;

[0022] The organic matter degradation rate So is obtained by the following formula:

[0023] ;

[0024] In the formula, dO / dt represents the change in dissolved oxygen in the water per unit time, and k represents the organic matter degradation rate constant;

[0025] The environmental data acquisition unit collects environmental data through solar radiation sensors and multispectral remote sensing instruments, including the influence factor Sw of solar radiation on water temperature changes and the complexity of algal community structure Sa.

[0026] The influence factor Sw on water temperature changes caused by sunlight is obtained through the following formula:

[0027] ;

[0028] In the formula, S1 represents the solar radiation absorption coefficient, I represents the solar radiation intensity incident on the water surface, S2 represents the light attenuation coefficient of the water body, zh represents the water depth, and e represents a constant.

[0029] The structural complexity Sa of an algal community is obtained using the following formula:

[0030] ;

[0031] In the formula, N represents the total number of algal community species, Pi represents the proportion of individuals of the i-th type of algae, and ln represents the logarithmic function with base e;

[0032] The acquired water quality and environmental data are combined and fitted into the original data set W.

[0033] Preferably, the data preprocessing module includes an anomaly detection unit and a data interpolation and standardization unit;

[0034] The anomaly detection unit removes outliers from the original data set W by using the Mahalanobis distance algorithm.

[0035] Data interpolation and standardization units use time series interpolation to complete the missing data in the original data set W, and use the min-max normalization method to map the data in the original data set W to the range [0,1] to obtain the water quality dataset SZW;

[0036] The water quality dataset SZW is obtained using the following formula:

[0037] ;

[0038] In the formula, Wd represents the d-th data item in the original data set W, Wdmin represents the valley value of the d-th data item in the original data set W, Wdmax represents the peak value of the d-th data item in the original data set W, and SZWd represents the d-th data item in the water quality dataset SZW.

[0039] Preferably, the feature engineering module includes a key feature extraction unit and a feature dimensionality reduction unit;

[0040] The key feature extraction unit extracts feature variables from the water quality dataset SZW, including the water activity index WAI and the pollution accumulation coefficient PAC. The water activity index WAI is used to determine the water activity status, and the pollution accumulation coefficient PAC is used to determine the water pollution status.

[0041] The water activity index (WAI) is obtained using the following formula:

[0042] ;

[0043] In the formula, maxSm represents the maximum value of the microturbulence intensity Sm, maxSo represents the maximum value of the organic matter degradation rate So, and maxSw represents the maximum value of the solar radiation effect factor Sw on water temperature change. These represent the preset weight values ​​for the microturbulence intensity Sm, the organic matter degradation rate So, and the influence factor Sw of solar radiation on water temperature change, respectively.

[0044] The active state of water bodies is obtained through the following methods:

[0045] When 0.7 < Water Activity Index (WAI) < 1.0, it indicates that the water body has strong activity and self-purification function.

[0046] When 0.4 ≤ Water Activity Index (WAI) ≤ 0.7, it indicates that the water activity is normal and can recover after being polluted.

[0047] When the water activity index (WAI) is less than 0.4, it indicates that the water activity is abnormal and the pollution is serious.

[0048] The Pollution Accumulation Factor (PAC) is obtained using the following formula:

[0049] ;

[0050] In the formula, maxSa represents the maximum value of algal community structure complexity Sa, and maxSr represents the maximum value of sediment suspension rate Sr. These represent the preset weight values ​​for algal community structure complexity Sa and sediment suspension rate Sr, respectively.

[0051] The pollution status of water bodies is obtained through the following methods:

[0052] When 0.8 < pollution accumulation coefficient PAC < 1.0, it indicates that pollutants have accumulated severely and the water body is in a eutrophication risk zone.

[0053] When 0.5 ≤ Pollution Accumulation Coefficient (PAC) ≤ 0.8, it indicates that pollutant accumulation is normal and regular monitoring is necessary.

[0054] When the pollution accumulation coefficient (PAC) is less than 0.5, it indicates that there is no accumulation of pollutants and the water quality is good.

[0055] Preferably, the feature dimensionality reduction unit uses principal component analysis to reduce the dimensionality of the extracted water activity index WAI and pollution accumulation coefficient PAC to construct a water quality feature set FZ;

[0056] The water quality feature set FZ is obtained through the following steps:

[0057] S1. Calculate the covariance matrix C using the water activity index WAI and the pollution accumulation coefficient PAC;

[0058] ;

[0059] In the formula, TZ represents the transpose of the matrix, μWAI represents the mean of the water activity index, and μPAC represents the mean of the pollution accumulation coefficient.

[0060] S2. Calculate eigenvalues ​​and eigenvectors through eigenvalue decomposition;

[0061] CV = λV;

[0062] In the formula, λ represents the eigenvalue and V represents the corresponding eigenvector;

[0063] S3. Dimensionally reduce the water activity index WAI and the pollution accumulation coefficient PAC to obtain the water quality feature set FZ;

[0064] The water quality feature set FZ is obtained using the following formula:

[0065] ;

[0066] In the formula, Vo represents the matrix composed of the first o eigenvectors.

[0067] Preferably, the model training and dynamic optimization module includes a deep learning training unit and a model dynamic optimization unit;

[0068] The deep learning training unit inputs the acquired water quality feature set FZ into the long short-term memory network model and trains the model to obtain the water quality monitoring result WQS.

[0069] The water quality monitoring result WQS is obtained using the following formula:

[0070] ;

[0071] In the formula, WQS represents the water quality monitoring result, specifically the water quality score, f represents the prediction function of the long short-term memory network model, and θ represents the parameters of the long short-term memory network model.

[0072] Preferably, the model dynamic optimization unit adjusts the long short-term memory network model through incremental learning, changing the parameters θ of the long short-term memory network model;

[0073] When new water quality and environmental data are collected, the long short-term memory network model is incrementally trained, the parameter θ is adjusted, and the new parameter nθ is calculated and obtained.

[0074] The new parameter nθ is obtained through the following formula:

[0075] ;

[0076] In the formula, η represents the learning rate, Loss represents the loss value, and ∇Loss represents the partial derivative of the loss function with respect to the parameter θ.

[0077] The loss value is obtained using the following formula:

[0078] ;

[0079] In the formula, M represents the number of samples, WQSj represents the actual water quality score of the j-th sample, and yWQSj is the model-predicted water quality score of the j-th sample.

[0080] Preferably, the water quality prediction module uses a self-attention mechanism model to perform time series processing on the water quality monitoring results WQS to obtain future water quality trends;

[0081] Using a self-attention mechanism to identify the correlation between water quality monitoring results (WQS) at different times;

[0082] The formula for calculating the self-attention mechanism is as follows:

[0083] ;

[0084] In the formula, Att represents the self-attention function, Qa represents the query matrix, ka represents the key matrix, Va represents the value matrix, and dk represents the vector dimension. This indicates that the query matrix Qa and the key matrix Ka are performed as a dot product, and somax indicates the normalization operation.

[0085] By using a self-attention mechanism model, future water quality scores can be predicted based on water quality monitoring results (WQS).

[0086] The water quality prediction formula is as follows:

[0087] ;

[0088] In the formula, WQS(t+p) represents the predicted water quality score at time t+p, and WQS(t) represents the water quality score at time t. represents the parameters of the self-attention mechanism model, and h represents the prediction function of the self-attention mechanism model.

[0089] Preferably, the decision support module assesses the water quality change trend based on the water quality score WQS(t+p) at time t+p and calculates the water quality deterioration risk index WQI.

[0090] The Water Quality Risk Index (WQI) is obtained using the following formula:

[0091] ;

[0092] In the formula, e represents a constant, and B represents an adjustment parameter;

[0093] The water quality warning level is obtained through the Water Quality Risk Index (WQI).

[0094] The formula for classifying water quality warning levels is as follows:

[0095] ;

[0096] The recommended course of action for low-risk situations is: No action required.

[0097] The decision-making recommendations for medium-risk areas are: water quality is affected by pollution, monitor it regularly; and take minor intervention measures, such as increasing oxygenation or water flow.

[0098] High-risk decision-making recommendations include: severely deteriorated water quality requiring remediation measures, such as pollution source tracing and chemical or biological remediation.

[0099] A water quality monitoring method for environmental protection includes the following steps:

[0100] Step 1: The data acquisition module collects water quality and environmental data through sensors and fits them into a raw data set W;

[0101] Step 2: The data preprocessing module performs anomaly detection, data interpolation, and standardization on the collected raw data set W to obtain the water quality dataset SZW;

[0102] Step 3: The feature engineering module extracts feature variables from the acquired water quality dataset SZW and uses principal component analysis to reduce the dimensionality of the features to obtain the water quality feature set FZ.

[0103] Step 4: The model training and dynamic optimization module uses the acquired water quality feature set FZ to predict water quality and obtain the water quality monitoring result WQS by using a long short-term memory network model.

[0104] Step 5: The water quality prediction module uses a self-attention mechanism model to perform time series prediction on the water quality monitoring results WQS, predicts the future trend of water quality, and obtains the water quality score WQS(t+p) at time t+p.

[0105] Step 6: The decision support module calculates the water quality deterioration risk index WQI based on the water quality score WQS(t+p) at time t+p and generates decision recommendations.

[0106] This invention provides a water quality monitoring system and method for environmental protection, which has the following beneficial effects:

[0107] (1) During system operation, water quality data and environmental data are combined to form a multi-dimensional data fusion analysis, avoiding the limitations of single parameter analysis and enabling the system to have the ability to judge global water quality changes. By combining the acquired water quality data and environmental data, a raw data set W is fitted, and optimization is performed at the data level to make the data more consistent and complete, reduce the impact of missing data on subsequent analysis, and improve data stability.

[0108] Monitoring the complexity of algal community structure allows for precise assessment of eutrophication levels in water bodies, providing a basis for accurate regulation in water quality management, such as adjusting the aquatic ecological balance and preventing algal blooms, thereby improving the efficiency of water environment protection. Employing high-precision sensors and remote sensing technology enables automated, remote, and continuous collection of water quality data, reducing manual intervention, lowering monitoring costs, and improving the efficiency and real-time nature of data acquisition, laying the foundation for smart water management and automated environmental management.

[0109] (2) By using time series interpolation to fill in missing data, the water quality data is made continuous in the time dimension, preventing monitoring blind spots caused by data loss and improving the integrity and stability of the system. The min-max normalization method is used to transform water quality data of different physical quantities to a uniform scale, enabling the deep learning model to converge faster, reducing the problem of uneven weight between features, and improving the computational efficiency and generalization ability of the water quality prediction model.

[0110] By calculating the Water Activity Index (WAI), we can scientifically assess the self-purification capacity of water bodies and identify whether they possess the ability to recover after pollution, providing more precise support for ecosystem health assessment. The Pollution Accumulation Coefficient (PAC) is used to assess the long-term accumulation of pollution in water bodies, distinguishing their pollution states and helping managers formulate scientifically sound water remediation measures to prevent the worsening of eutrophication.

[0111] (3) Using a Long Short-Term Memory (LSTM) network to process water quality characteristic data can effectively capture the time dependence of water quality changes, learn the potential patterns in historical data, and improve the accuracy of water quality monitoring results. Through deep learning training, the system can automatically adapt to the characteristics of different water bodies, avoiding the limitations of traditional fixed-rule methods, and making water quality scoring more accurate and reliable. Through incremental learning, the model is continuously optimized, enabling it to adjust its parameters as new water quality and environmental data are added, thereby improving the model's adaptability to long-term water quality trends.

[0112] The gradient descent optimization algorithm is employed to dynamically adjust model parameters by calculating the loss value, making the training process more efficient, reducing computational resource consumption, accelerating model convergence, and improving the real-time response capability of the water quality monitoring system. Through the self-learning capability of neural networks, the system can extract key patterns from historical water quality data without the need for manually setting complex rules, reducing human intervention, increasing automation, and making water environment monitoring more intelligent.

[0113] (4) Employing multi-sensor data fusion ensures the accuracy of data acquisition, reduces data deviation caused by single sensor errors, and improves the reliability of water quality monitoring. Through anomaly detection and data interpolation techniques, erroneous values ​​are eliminated and missing data is supplemented, ensuring the integrity of water quality data over time and guaranteeing continuous, stable, and usable monitoring data. Standardized methods are used to process data, maintaining consistent scale across different physical quantities, improving the standardization of data input, and making subsequent model training more efficient. The Water Quality Risk Index (WQI) is used to quantify water quality change trends, providing a more scientific method for assessing water quality health, avoiding errors from subjective human judgment, and improving the accuracy of water quality early warning.

[0114] By analyzing water quality scores, trends of water quality deterioration can be identified in advance, and intelligent decision-making suggestions at different risk levels can be generated to ensure that management departments can take targeted water quality management measures and optimize water resource management efficiency. Attached Figure Description

[0115] Figure 1 This is a schematic flowchart of a water quality monitoring system for environmental protection according to the present invention.

[0116] Figure 2 This is a schematic diagram of the steps of a water quality monitoring method for environmental protection according to the present invention;

[0117] Figure 3 This is a schematic diagram of the water quality deterioration risk assessment process of the present invention;

[0118] Figure 4 This is a line graph showing the change in the water activity index according to the present invention.

[0119] Figure 5 This is a bar chart of the pollution accumulation index of this invention. Detailed Implementation

[0120] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0121] Example 1

[0122] This invention provides a water quality monitoring system for environmental protection. Please refer to [link / reference]. Figures 1-5 It includes a data acquisition module, a data preprocessing module, a feature engineering module, a model training and dynamic optimization module, a water quality prediction module, and a decision support module;

[0123] The data acquisition module collects water quality and environmental data through sensors and fits them into a raw data set W;

[0124] The data preprocessing module performs anomaly detection, data interpolation, and standardization on the collected raw data set W to obtain the water quality dataset SZW;

[0125] The feature engineering module extracts feature variables from the acquired water quality dataset SZW and uses principal component analysis to reduce the dimensionality of the features to obtain the water quality feature set FZ.

[0126] The model training and dynamic optimization module uses the acquired water quality feature set FZ to predict water quality and obtain the water quality monitoring result WQS by using a long short time memory network model.

[0127] The water quality prediction module uses a self-attention mechanism model to perform time series prediction on the water quality monitoring results WQS, predicts the future trend of water quality, and obtains the water quality score WQS(t+p) at time t+p.

[0128] The decision support module calculates the water quality deterioration risk index WQI based on the water quality score WQS(t+p) at time t+p and generates decision recommendations.

[0129] In this embodiment, a deep learning model combined with a self-attention mechanism is used for water quality prediction. This can dynamically analyze water quality change trends, avoid false alarms and missed alarms caused by fixed thresholds, and achieve more intelligent water quality monitoring.

[0130] By employing Long Short-Term Memory (LSTM) networks for water quality prediction and combining this with self-attention mechanisms for future trend analysis, the system can predict changes in water quality scores in advance. This allows management departments to take measures before water quality deteriorates, reducing the risk of water pollution spread and improving the scientific rigor and timeliness of early warnings. A data preprocessing module is used for outlier detection, data interpolation, and standardization to ensure data integrity and consistency, reduce monitoring biases caused by sensor errors or data loss, and improve data quality, thereby enhancing the overall system's monitoring accuracy and stability.

[0131] Feature engineering is used to extract characteristic variables, and principal component analysis is employed for dimensionality reduction to select the key variables that best reflect water quality changes, improving the model's computational efficiency and enabling the system to adapt to the water quality characteristics of different water areas, thus enhancing its generalization ability. A model training and dynamic optimization module is used to incrementally learn and optimize model parameters as water quality data is updated, ensuring the system can continuously update its predictive capabilities in response to changes in the water environment, avoiding model aging and improving the reliability of long-term predictions. The decision support module calculates the Water Quality Indicator (WQI) and generates targeted governance recommendations, enabling the recommendation of appropriate water environment governance measures based on different water quality risk levels, improving the scientific basis of management decisions and enhancing the efficiency of water resource management.

[0132] Example 2

[0133] This embodiment is an explanation based on Embodiment 1. Please refer to it. Figure 1 Specifically: the data acquisition module includes a water quality data acquisition unit and an environmental data acquisition unit;

[0134] The water quality data acquisition unit collects water quality data, including microturbulence intensity Sm, sediment suspension rate Sr, and organic matter degradation rate So, through a microturbulence intensity sensor, a suspended particle optical sensor, and a dissolved oxygen and organic matter analyzer.

[0135] The microturbulence intensity Sm was collected by a microturbulence intensity sensor, the sediment suspension rate Sr was obtained by a suspended particle optical sensor, and the organic matter degradation rate So was collected by a dissolved oxygen and organic matter analyzer.

[0136] The microturbulence intensity Sm is obtained by the following formula:

[0137] ;

[0138] In the formula, xus, yus and zus represent the instantaneous flow velocities of the water in the x, y and z directions, respectively, and xup, yup and zup represent the average flow velocities of the water in the x, y and z directions, respectively.

[0139] Sediment suspension rate Sr is obtained using the following formula:

[0140] ;

[0141] In the formula, Ct represents the concentration of suspended particles in the collected water sample, and Co represents the concentration of suspended particles in the background water body;

[0142] The organic matter degradation rate So is obtained by the following formula:

[0143] ;

[0144] In the formula, dO / dt represents the change in dissolved oxygen in the water per unit time, and k represents the organic matter degradation rate constant;

[0145] The environmental data acquisition unit collects environmental data through solar radiation sensors and multispectral remote sensing instruments, including the influence factor Sw of solar radiation on water temperature changes and the complexity of algal community structure Sa.

[0146] The influence factor Sw on water temperature changes caused by sunlight is obtained through the following formula:

[0147] ;

[0148] In the formula, S1 represents the solar radiation absorption coefficient, I represents the solar radiation intensity incident on the water surface, S2 represents the light attenuation coefficient of the water body, zh represents the water depth, and e represents a constant.

[0149] The structural complexity Sa of an algal community is obtained using the following formula:

[0150] ;

[0151] In the formula, N represents the total number of algal community species, Pi represents the proportion of individuals of the i-th type of algae, and ln represents the logarithmic function with base e;

[0152] The acquired water quality and environmental data are combined and fitted into the original data set W.

[0153] In this embodiment, high-precision equipment such as microturbulence intensity sensors, suspended particle optical sensors, and dissolved oxygen and organic matter analyzers are used to accurately measure water flow, suspended particle distribution, and organic pollution degradation, reducing data acquisition errors and providing high-quality basic data for subsequent analysis. Solar radiation sensors and multispectral remote sensing instruments are employed to achieve real-time monitoring of environmental factors such as solar radiation intensity, water temperature changes, and algal community structure, enabling the system to comprehensively assess the impact of the external environment on water quality and improve the adaptability of the prediction model.

[0154] By combining water quality and environmental data, a multi-dimensional data fusion analysis is formed, avoiding the limitations of single-parameter analysis and enabling the system to judge global water quality changes. The acquired water quality and environmental data are combined and fitted into an original data set W. Optimization at the data level makes the data more consistent and complete, reduces the impact of missing data on subsequent analysis, and improves data stability.

[0155] Monitoring the complexity of algal community structure allows for precise assessment of eutrophication levels in water bodies, providing a basis for accurate regulation in water quality management, such as adjusting the aquatic ecological balance and preventing algal blooms, thereby improving the efficiency of water environment protection. Employing high-precision sensors and remote sensing technology enables automated, remote, and continuous collection of water quality data, reducing manual intervention, lowering monitoring costs, and improving the efficiency and real-time nature of data acquisition, laying the foundation for smart water management and automated environmental management.

[0156] Example 3

[0157] This embodiment is an explanation based on Embodiment 2. Please refer to it. Figure 1 , Figure 4 and Figure 5 Specifically: the data preprocessing module includes an anomaly detection unit and a data interpolation and standardization unit;

[0158] The anomaly detection unit removes outliers from the original data set W by using the Mahalanobis distance algorithm.

[0159] Data interpolation and standardization units use time series interpolation to complete the missing data in the original data set W, and use the min-max normalization method to map the data in the original data set W to the range [0,1] to obtain the water quality dataset SZW;

[0160] The water quality dataset SZW is obtained using the following formula:

[0161] ;

[0162] In the formula, Wd represents the d-th data item in the original data set W, Wdmin represents the valley value of the d-th data item in the original data set W, Wdmax represents the peak value of the d-th data item in the original data set W, and SZWd represents the d-th data item in the water quality dataset SZW.

[0163] The feature engineering module includes a key feature extraction unit and a feature dimensionality reduction unit;

[0164] The key feature extraction unit extracts feature variables from the water quality dataset SZW, including the water activity index WAI and the pollution accumulation coefficient PAC. The water activity index WAI is used to determine the water activity status, and the pollution accumulation coefficient PAC is used to determine the water pollution status.

[0165] The water activity index (WAI) is obtained using the following formula:

[0166] ;

[0167] In the formula, maxSm represents the maximum value of the microturbulence intensity Sm, maxSo represents the maximum value of the organic matter degradation rate So, and maxSw represents the maximum value of the solar radiation effect factor Sw on water temperature change. These represent the preset weight values ​​for the microturbulence intensity Sm, the organic matter degradation rate So, and the influence factor Sw of solar radiation on water temperature change, respectively.

[0168] Specific examples:

[0169] Preset They are 0.4, 0.3, and 0.3 respectively;

[0170] The maximum value of the microturbulence intensity Sm is maxSm=1.0;

[0171] The maximum value of the organic matter degradation rate So is maxSo = 0.8;

[0172] The maximum value of the influence factor Sw on water temperature change caused by sunlight is maxSw = 1.0;

[0173] The microturbulence intensity Sm = 0.17, the organic matter degradation rate So = 0.25, and the influence factor of solar radiation on water temperature change Sw = 0.37 were obtained.

[0174] Calculate and obtain the water activity index WAI:

[0175] ;

[0176] Table 1. Data for calculating the water body activity index:

[0177] Group number Microturbulence intensity Sm Organic matter degradation rate So Factors influencing water temperature changes: Sw Water Activity Index (WAI) Group 1 0.17 0.25 0.37 0.272 Group 2 0.60 0.63 0.85 0.731 Group 3 0.46 0.76 0.54 0.631 Group 4 0.54 0.60 0.48 0.585

[0178] The active state of water bodies is obtained through the following methods:

[0179] When 0.7 < Water Activity Index (WAI) < 1.0, it indicates that the water body has strong activity and self-purification function.

[0180] When 0.4 ≤ Water Activity Index (WAI) ≤ 0.7, it indicates that the water activity is normal and can recover after being polluted.

[0181] When the water activity index (WAI) is less than 0.4, it indicates that the water activity is abnormal and the pollution is serious.

[0182] The Pollution Accumulation Factor (PAC) is obtained using the following formula:

[0183] ;

[0184] In the formula, maxSa represents the maximum value of algal community structure complexity Sa, and maxSr represents the maximum value of sediment suspension rate Sr. These represent the preset weight values ​​for algal community structure complexity Sa and sediment suspension rate Sr, respectively.

[0185] Specific examples:

[0186] Preset They are 0.6 and 0.4 respectively;

[0187] The maximum value of algal community structural complexity Sa is maxSa = 0.9;

[0188] The maximum value of sediment suspension rate Sr is maxSr = 0.8;

[0189] The algal community structure complexity Sa = 0.85 and the sediment suspension rate Sr = 0.62 were obtained.

[0190] Calculate and obtain the pollution accumulation factor (PAC):

[0191] ;

[0192] Table 2: Data for calculating pollution cumulative coefficients

[0193] Group number Algal community structural complexity Sa Sediment suspension rate Sr Pollution Accumulation Coefficient (PAC) Group 1 0.85 0.62 0.876 Group 2 0.29 0.47 0.428 Group 3 0.59 0.78 0.783 Group 4 0.58 0.28 0.526

[0194] The pollution status of water bodies is obtained through the following methods:

[0195] When 0.8 < pollution accumulation coefficient PAC < 1.0, it indicates that pollutants have accumulated severely and the water body is in a eutrophication risk zone.

[0196] When 0.5 ≤ Pollution Accumulation Coefficient (PAC) ≤ 0.8, it indicates that pollutant accumulation is normal and regular monitoring is necessary.

[0197] When the pollution accumulation coefficient (PAC) is less than 0.5, it indicates that there is no accumulation of pollutants and the water quality is good.

[0198] The feature dimensionality reduction unit uses principal component analysis to reduce the dimensionality of the extracted water activity index WAI and pollution accumulation coefficient PAC, and constructs the water quality feature set FZ.

[0199] The water quality feature set FZ is obtained through the following steps:

[0200] S1. Calculate the covariance matrix C using the water activity index WAI and the pollution accumulation coefficient PAC;

[0201] ;

[0202] In the formula, TZ represents the transpose of the matrix, μWAI represents the mean of the water activity index, and μPAC represents the mean of the pollution accumulation coefficient.

[0203] S2. Calculate eigenvalues ​​and eigenvectors through eigenvalue decomposition;

[0204] CV = λV;

[0205] In the formula, λ represents the eigenvalue and V represents the corresponding eigenvector;

[0206] S3. Dimensionally reduce the water activity index WAI and the pollution accumulation coefficient PAC to obtain the water quality feature set FZ;

[0207] The water quality feature set FZ is obtained using the following formula:

[0208] ;

[0209] In the formula, Vo represents the matrix composed of the first o eigenvectors.

[0210] In this embodiment, the Mahalanobis distance algorithm is used for anomaly detection, effectively removing erroneous and abnormal data from the original data, avoiding the impact of data anomalies on subsequent analysis, and improving the data reliability and accuracy of the water quality monitoring system. Time series interpolation is used to complete missing data, ensuring the continuity of water quality data over time, preventing monitoring blind spots caused by data loss, and improving the system's integrity and stability. The min-max normalization method is employed to transform water quality data of different physical quantities to a uniform scale, enabling the deep learning model to converge faster, reducing the problem of uneven weight distribution among features, and improving the computational efficiency and generalization ability of the water quality prediction model.

[0211] By calculating the Water Activity Index (WAI), the self-purification capacity of water bodies can be scientifically assessed, and their ability to recover after pollution can be identified, providing more precise support for ecosystem health assessment. The Pollution Accumulation Coefficient (PAC) is used to assess the long-term accumulation of pollution in water bodies, distinguishing their pollution states and helping managers formulate scientifically sound water remediation measures to prevent the worsening of eutrophication. Based on the dynamic trends of the WAI and PAC, the system can intelligently classify water pollution states, providing management departments with more accurate pollution early warning information and preventing ecological upheavals caused by pollution accumulation. The classification of water activity and pollution accumulation states provides a more intuitive view of water quality, enabling decision-makers to quickly identify pollution risks and take corresponding remediation measures, thus improving the efficiency of water environment management.

[0212] Principal component analysis was used to reduce the dimensionality of the water activity index (WAI) and the pollution accumulation coefficient (PAC), remove redundant features, and reduce computational complexity, making the water quality monitoring system more efficient when processing large-scale data.

[0213] By calculating the covariance matrix, performing eigenvalue decomposition, and dimensionality reduction, the main features of water quality changes are extracted, enabling the system to focus more on key variables and improve the accuracy of water quality predictions. Combined with a comprehensive analysis of water activity and pollution accumulation characteristics, the system can adapt to the characteristic changes of different water body types, improving the model's adaptability and making prediction results more accurate and reliable. The optimized water quality feature set FZ ensures more refined input data for subsequent deep learning models, improving the stability and intelligence of the water quality monitoring system and enabling it to maintain efficient operation over the long term. Through decision support based on water quality characteristics, the system can provide more scientific water environment governance solutions, reducing blind governance and improving the efficiency of pollution prevention and ecological restoration.

[0214] Example 4

[0215] This embodiment is an explanation based on Embodiment 3. Please refer to it. Figure 1 Specifically: the model training and dynamic optimization module includes a deep learning training unit and a model dynamic optimization unit;

[0216] The deep learning training unit inputs the acquired water quality feature set FZ into the long short-term memory network model and trains the model to obtain the water quality monitoring result WQS.

[0217] The water quality monitoring result WQS is obtained using the following formula:

[0218] ;

[0219] In the formula, WQS represents the water quality monitoring result, specifically the water quality score, f represents the prediction function of the long short-term memory network model, and θ represents the parameters of the long short-term memory network model.

[0220] The model dynamic optimization unit adjusts the Long Short-Term Memory (LSTM) network model through incremental learning, changing the parameters θ of the LSM network model.

[0221] When new water quality and environmental data are collected, the long short-term memory network model is incrementally trained, the parameter θ is adjusted, and the new parameter nθ is calculated and obtained.

[0222] The new parameter nθ is obtained through the following formula:

[0223] ;

[0224] In the formula, η represents the learning rate, Loss represents the loss value, and ∇Loss represents the partial derivative of the loss function with respect to the parameter θ.

[0225] The loss value is obtained using the following formula:

[0226] ;

[0227] In the formula, M represents the number of samples, WQSj represents the actual water quality score of the j-th sample, and yWQSj is the model-predicted water quality score of the j-th sample.

[0228] In this embodiment, a Long Short-Term Memory (LSTM) network is used to process water quality characteristic data, effectively capturing the time dependence of water quality changes, learning potential patterns in historical data, and improving the accuracy of water quality monitoring results. Through deep learning training, the system can automatically adapt to the characteristics of different water bodies, avoiding the limitations of traditional fixed-rule methods and making water quality scoring more accurate and reliable. Incremental learning continuously optimizes the model, enabling it to adjust its parameters with the addition of new water quality and environmental data, improving the model's adaptability to long-term water quality trends.

[0229] The gradient descent optimization algorithm is employed to dynamically adjust model parameters by calculating the loss value, making the training process more efficient, reducing computational resource consumption, accelerating model convergence, and improving the real-time response capability of the water quality monitoring system. Through the self-learning capability of neural networks, the system can extract key patterns from historical water quality data without the need for manually setting complex rules, reducing human intervention, increasing automation, and making water environment monitoring more intelligent.

[0230] By incorporating the water quality feature set FZ, the model can autonomously learn comprehensive information such as water activity, pollution accumulation, and environmental changes, generating more reasonable water quality monitoring results and enhancing the overall system's intelligence. Employing a dynamic parameter update mechanism, the system can adjust the model instantly after data updates, improving the response speed to sudden water quality changes and shortening the time interval for water pollution identification and early warning. Through a scalable deep learning model, the system can be applied to different types of water bodies, such as rivers, lakes, reservoirs, and industrial wastewater treatment, demonstrating broader application value. Combined with accurate water quality scoring, the system can provide more scientific pollution prevention strategies, optimize water resource allocation, and improve the accuracy and operability of water quality management.

[0231] Example 5

[0232] This embodiment is an explanation based on Embodiment 4. Please refer to it. Figure 1 and Figure 3 Specifically: the water quality prediction module uses a self-attention mechanism model to perform time series processing on the water quality monitoring results WQS to obtain future water quality trends;

[0233] Using a self-attention mechanism to identify the correlation between water quality monitoring results (WQS) at different times;

[0234] The formula for calculating the self-attention mechanism is as follows:

[0235] ;

[0236] In the formula, Att represents the self-attention function, Qa represents the query matrix, ka represents the key matrix, Va represents the value matrix, and dk represents the vector dimension. This indicates that the query matrix Qa and the key matrix Ka are performed as a dot product, and somax indicates the normalization operation.

[0237] By using a self-attention mechanism model, future water quality scores can be predicted based on water quality monitoring results (WQS).

[0238] The water quality prediction formula is as follows:

[0239] ;

[0240] In the formula, WQS(t+p) represents the predicted water quality score at time t+p, and WQS(t) represents the water quality score at time t. represents the parameters of the self-attention mechanism model, and h represents the prediction function of the self-attention mechanism model.

[0241] The decision support module assesses the water quality change trend based on the water quality score WQS(t+p) at time t+p and calculates the water quality deterioration risk index WQI.

[0242] The Water Quality Risk Index (WQI) is obtained using the following formula:

[0243] ;

[0244] In the formula, e represents a constant, and B represents an adjustment parameter;

[0245] The water quality warning level is obtained through the Water Quality Risk Index (WQI).

[0246] The formula for classifying water quality warning levels is as follows:

[0247] ;

[0248] The recommended course of action for low-risk situations is: No action required.

[0249] The recommended course of action for medium-risk water pollution is: regular monitoring is necessary.

[0250] The high-risk decision recommendation is: water quality has deteriorated significantly, and remediation measures should be implemented.

[0251] In this embodiment, a self-attention mechanism model is used to process water quality monitoring results, which can accurately identify the correlation between different time steps and improve the accuracy of water quality trend prediction. Through deep time series analysis, the system can more accurately capture water quality change patterns, avoid the over-reliance on short-term fluctuations in traditional methods, and improve the prediction accuracy of future water quality scores. By identifying long-term and short-term water quality change trends through the self-attention mechanism, potential signals of water quality deterioration can be detected earlier, improving the response capability of the early warning system. Combined with the prediction of future water quality scores WQS(t+p), the system can provide early warnings of water quality changes several days to weeks before pollution occurs, enabling water resource managers to take measures in advance to reduce the impact of water pollution.

[0252] Using the Water Quality Index (WQI) for quantitative assessment provides a more scientific measure of water quality change trends, avoiding errors caused by subjective experience. Calculating the WQI using an exponential growth model rapidly amplifies risk assessment results when water quality scores decline, making the system more sensitive to potential water quality deterioration and improving the accuracy of early warnings. A tiered early warning mechanism classifies water quality changes into low, medium, and high risk levels, preventing false alarms during minor water quality fluctuations while ensuring rapid response in cases of severe pollution.

[0253] Different treatment plans are provided based on the water quality risk level, such as no intervention for low risk, regular monitoring for medium risk, and emergency measures for high risk, ensuring more precise and efficient water quality management. Combined with future water quality trend predictions, the system can help managers develop long-term pollution prevention strategies, optimize water resource utilization, and improve the sustainability of pollution control.

[0254] Employing a self-attention mechanism for parallel computing reduces reliance on traditional time series models, improving computational efficiency for water quality prediction and enabling the system to respond more quickly to changes in water quality data. By automatically updating the parameters of the water quality prediction model, manual intervention is reduced, allowing the water quality monitoring system to adapt to changes and improving predictive capabilities and long-term stability.

[0255] Example 6

[0256] A water quality monitoring method for environmental protection, please refer to... Figure 2 Specifically, it includes the following steps:

[0257] Step 1: The data acquisition module collects water quality and environmental data through sensors and fits them into a raw data set W;

[0258] Step 2: The data preprocessing module performs anomaly detection, data interpolation, and standardization on the collected raw data set W to obtain the water quality dataset SZW;

[0259] Step 3: The feature engineering module extracts feature variables from the acquired water quality dataset SZW and uses principal component analysis to reduce the dimensionality of the features to obtain the water quality feature set FZ.

[0260] Step 4: The model training and dynamic optimization module uses the acquired water quality feature set FZ to predict water quality and obtain the water quality monitoring result WQS by using a long short-term memory network model.

[0261] Step 5: The water quality prediction module uses a self-attention mechanism model to perform time series prediction on the water quality monitoring results WQS, predicts the future trend of water quality, and obtains the water quality score WQS(t+p) at time t+p.

[0262] Step 6: The decision support module calculates the water quality deterioration risk index WQI based on the water quality score WQS(t+p) at time t+p and generates decision recommendations.

[0263] In this embodiment, multi-sensor data fusion is employed to ensure the accuracy of data acquisition, reduce data deviations caused by errors in single sensors, and improve the reliability of water quality monitoring. Anomaly detection and data interpolation techniques are used to remove erroneous values ​​and fill in missing data, ensuring the integrity of water quality data over time and guaranteeing continuous, stable, and usable monitoring data. Standardized methods are used to process the data, maintaining consistent scale across different physical quantities, improving the standardization of data input, and making subsequent model training more efficient. The Water Quality Index (WQI) is used to quantify water quality change trends, providing a more scientific method for assessing water quality health, avoiding errors from subjective human judgment, and improving the accuracy of water quality early warnings.

[0264] By analyzing water quality scores and trends, we can identify water quality deterioration trends in advance and generate intelligent decision-making suggestions at different risk levels. This ensures that management departments can take targeted water quality control measures and optimize water resource management efficiency. Through these intelligent decision-making suggestions, managers can proactively deploy pollution control measures, such as optimizing water flow, adjusting water quality parameters, and controlling pollutant sources, reducing the cost and time of water pollution control and improving the sustainability of water environmental protection.

[0265] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A water quality monitoring system for environmental protection, characterized in that: It includes a data acquisition module, a data preprocessing module, a feature engineering module, a model training and dynamic optimization module, a water quality prediction module, and a decision support module; The data acquisition module collects water quality and environmental data through sensors and fits them into a raw data set W; The data preprocessing module performs anomaly detection, data interpolation, and standardization on the collected raw data set W to obtain the water quality dataset SZW; The feature engineering module extracts feature variables from the acquired water quality dataset SZW and uses principal component analysis to reduce the dimensionality of the features to obtain the water quality feature set FZ. The model training and dynamic optimization module uses the acquired water quality feature set FZ to predict water quality and obtain the water quality monitoring result WQS by using a long short time memory network model. The water quality prediction module uses a self-attention mechanism model to perform time series prediction on the water quality monitoring results WQS, predicts the future trend of water quality, and obtains the water quality score WQS(t+p) at time t+p. The decision support module calculates the water quality deterioration risk index WQI based on the water quality score WQS(t+p) at time t+p and generates decision recommendations.

2. The water quality monitoring system for environmental protection according to claim 1, characterized in that: The data acquisition module includes a water quality data acquisition unit and an environmental data acquisition unit; The water quality data acquisition unit collects water quality data, including microturbulence intensity Sm, sediment suspension rate Sr, and organic matter degradation rate So, through a microturbulence intensity sensor, a suspended particle optical sensor, and a dissolved oxygen and organic matter analyzer. The microturbulence intensity Sm was collected by a microturbulence intensity sensor, the sediment suspension rate Sr was obtained by a suspended particle optical sensor, and the organic matter degradation rate So was collected by a dissolved oxygen and organic matter analyzer. The microturbulence intensity Sm is obtained by the following formula: ； In the formula, xus, yus and zus represent the instantaneous flow velocities of the water in the x, y and z directions, respectively, and xup, yup and zup represent the average flow velocities of the water in the x, y and z directions, respectively. Sediment suspension rate Sr is obtained using the following formula: ； In the formula, Ct represents the concentration of suspended particles in the collected water sample, and Co represents the concentration of suspended particles in the background water body; The organic matter degradation rate So is obtained by the following formula: ； In the formula, dO / dt represents the change in dissolved oxygen in the water per unit time, and k represents the organic matter degradation rate constant; The environmental data acquisition unit collects environmental data through solar radiation sensors and multispectral remote sensing instruments, including the influence factor Sw of solar radiation on water temperature changes and the complexity of algal community structure Sa. The influence factor Sw on water temperature changes caused by sunlight is obtained through the following formula: ； In the formula, S1 represents the solar radiation absorption coefficient, I represents the solar radiation intensity incident on the water surface, S2 represents the light attenuation coefficient of the water body, zh represents the water depth, and e represents a constant. The structural complexity Sa of an algal community is obtained using the following formula: ； In the formula, N represents the total number of algal community species, Pi represents the proportion of individuals of the i-th type of algae, and ln represents the logarithmic function with base e; The acquired water quality and environmental data are combined and fitted into the original data set W.

3. A water quality monitoring system for environmental protection according to claim 1, characterized in that: The data preprocessing module includes an anomaly detection unit and a data interpolation and standardization unit; The anomaly detection unit removes outliers from the original data set W by using the Mahalanobis distance algorithm. Data interpolation and standardization units use time series interpolation to complete the missing data in the original data set W, and use the min-max normalization method to map the data in the original data set W to the range [0,1] to obtain the water quality dataset SZW; The water quality dataset SZW is obtained using the following formula: ； In the formula, Wd represents the d-th data item in the original data set W, Wdmin represents the valley value of the d-th data item in the original data set W, Wdmax represents the peak value of the d-th data item in the original data set W, and SZWd represents the d-th data item in the water quality dataset SZW.

4. A water quality monitoring system for environmental protection according to claim 1, characterized in that: The feature engineering module includes a key feature extraction unit and a feature dimensionality reduction unit; The key feature extraction unit extracts feature variables from the water quality dataset SZW, including the water activity index WAI and the pollution accumulation coefficient PAC. The water activity index WAI is used to determine the water activity status, and the pollution accumulation coefficient PAC is used to determine the water pollution status. The water activity index (WAI) is obtained using the following formula: ； In the formula, maxSm represents the maximum value of the microturbulence intensity Sm, maxSo represents the maximum value of the organic matter degradation rate So, and maxSw represents the maximum value of the solar radiation effect factor Sw on water temperature change. These represent the preset weight values ​​for the microturbulence intensity Sm, the organic matter degradation rate So, and the influence factor Sw of solar radiation on water temperature change, respectively. The active state of water bodies is obtained through the following methods: When 0.7 < Water Activity Index (WAI) < 1.0, it indicates that the water body has strong activity and self-purification function. When 0.4 ≤ Water Activity Index (WAI) ≤ 0.7, it indicates that the water activity is normal and can recover after being polluted. When the water activity index (WAI) is less than 0.4, it indicates that the water activity is abnormal and the pollution is serious. The Pollution Accumulation Factor (PAC) is obtained using the following formula: ； In the formula, maxSa represents the maximum value of algal community structure complexity Sa, and maxSr represents the maximum value of sediment suspension rate Sr. These represent the preset weight values ​​for algal community structure complexity Sa and sediment suspension rate Sr, respectively. The pollution status of water bodies is obtained through the following methods: When 0.8 < pollution accumulation coefficient PAC < 1.0, it indicates that pollutants have accumulated severely and the water body is in a eutrophication risk zone. When 0.5 ≤ Pollution Accumulation Coefficient (PAC) ≤ 0.8, it indicates that pollutant accumulation is normal and regular monitoring is necessary. When the pollution accumulation coefficient (PAC) is less than 0.5, it indicates that there is no accumulation of pollutants and the water quality is good.

5. A water quality monitoring system for environmental protection according to claim 4, characterized in that: The feature dimensionality reduction unit uses principal component analysis to reduce the dimensionality of the extracted water activity index WAI and pollution accumulation coefficient PAC, and constructs the water quality feature set FZ. The water quality feature set FZ is obtained through the following steps: S1. Calculate the covariance matrix C using the water activity index WAI and the pollution accumulation coefficient PAC; ； In the formula, TZ represents the transpose of the matrix, μWAI represents the mean of the water activity index, and μPAC represents the mean of the pollution accumulation coefficient. S2. Calculate eigenvalues ​​and eigenvectors through eigenvalue decomposition; CV = λV; In the formula, λ represents the eigenvalue and V represents the corresponding eigenvector; S3. Dimensionally reduce the water activity index WAI and the pollution accumulation coefficient PAC to obtain the water quality feature set FZ; The water quality feature set FZ is obtained using the following formula: ； In the formula, Vo represents the matrix composed of the first o eigenvectors.

6. A water quality monitoring system for environmental protection according to claim 1, characterized in that: The model training and dynamic optimization module includes a deep learning training unit and a model dynamic optimization unit; The deep learning training unit inputs the acquired water quality feature set FZ into the long short-term memory network model and trains the model to obtain the water quality monitoring result WQS. Water quality monitoring results (WQS) are obtained using the following formula: ； In the formula, WQS represents the water quality monitoring result, specifically the water quality score, f represents the prediction function of the long short-term memory network model, and θ represents the parameters of the long short-term memory network model.

7. A water quality monitoring system for environmental protection according to claim 6, characterized in that: The model dynamic optimization unit adjusts the Long Short-Term Memory (LSTM) network model through incremental learning, changing the parameters θ of the LSM network model. When new water quality and environmental data are collected, the long short-term memory network model is incrementally trained, the parameter θ is adjusted, and the new parameter nθ is calculated and obtained. The new parameter nθ is obtained using the following formula: ； In the formula, η represents the learning rate, Loss represents the loss value, and ∇Loss represents the partial derivative of the loss function with respect to the parameter θ. The loss value is obtained using the following formula: ； In the formula, M represents the number of samples, WQSj represents the actual water quality score of the j-th sample, and yWQSj is the model-predicted water quality score of the j-th sample.

8. A water quality monitoring system for environmental protection according to claim 1, characterized in that: The water quality prediction module uses a self-attention mechanism model to perform time series processing on the water quality monitoring results WQS to obtain future water quality trends. Using a self-attention mechanism to identify the correlation between water quality monitoring results (WQS) at different times; The formula for calculating the self-attention mechanism is as follows: ； In the formula, Att represents the self-attention function, Qa represents the query matrix, ka represents the key matrix, Va represents the value matrix, and dk represents the vector dimension. This indicates that the query matrix Qa and the key matrix Ka are performed as a dot product, and somax indicates the normalization operation. By using a self-attention mechanism model, future water quality scores can be predicted based on water quality monitoring results (WQS). The water quality prediction formula is as follows: ； In the formula, WQS(t+p) represents the predicted water quality score at time t+p, and WQS(t) represents the water quality score at time t. represents the parameters of the self-attention mechanism model, and h represents the prediction function of the self-attention mechanism model.

9. A water quality monitoring system for environmental protection according to claim 1, characterized in that: The decision support module assesses the water quality change trend based on the water quality score WQS(t+p) at time t+p and calculates the water quality deterioration risk index WQI. The Water Quality Risk Index (WQI) is obtained using the following formula: ； In the formula, e represents a constant, and B represents an adjustment parameter; The water quality warning level is obtained through the Water Quality Risk Index (WQI). The formula for classifying water quality warning levels is as follows: ； The recommended course of action for low-risk situations is: No action required. The recommended course of action for medium-risk water pollution is: regular monitoring is necessary. The high-risk decision recommendation is: water quality has deteriorated significantly, and remediation measures should be implemented.

10. A water quality monitoring method for environmental protection, applied to a water quality monitoring system for environmental protection as described in any one of claims 1 to 9, characterized in that: Includes the following steps: Step 1: The data acquisition module collects water quality and environmental data through sensors and fits them into a raw data set W; Step 2: The data preprocessing module performs anomaly detection, data interpolation, and standardization on the collected raw data set W to obtain the water quality dataset SZW; Step 3: The feature engineering module extracts feature variables from the acquired water quality dataset SZW and uses principal component analysis to reduce the dimensionality of the features to obtain the water quality feature set FZ. Step 4: The model training and dynamic optimization module uses the acquired water quality feature set FZ to predict water quality and obtain the water quality monitoring result WQS by using a long short-term memory network model. Step 5: The water quality prediction module uses a self-attention mechanism model to perform time series prediction on the water quality monitoring results WQS, predicts the future trend of water quality, and obtains the water quality score WQS(t+p) at time t+p. Step 6: The decision support module calculates the water quality deterioration risk index WQI based on the water quality score WQS(t+p) at time t+p and generates decision recommendations.

Images